APPLIED ELECTROCHEMISTRY THE MACMILLAN COMPANY NEW YORK BOSTON CHICAGO SAN FRANCISCO MACMILLAN & CO., LIMITED LONDON BOMBAY CALCUTTA MELBOURNE THE MACMILLAN CO. OF CANADA, LTD. TORONTO APPLIED ELECTROCHEMISTRY BY M. DEKAY THOMPSON, PH.D. h ASSISTANT PROFESSOR OF ELECTROCHEMISTRY IN THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY gorfc THE MACMILLAN COMPANY 1914 All rights reserved COPYBIGHT, 1911, BT THE MACMILLAN COMPANY. Set up and electrotyped. Published May, 1911. Reprinted April, 1914. Nortooot) J. 8. Gushing Co. Berwick & Smith Co. Norwood, Mass., U.S.A. PREFACE THE following book was written to supply a need felt by the author in giving a course of lectures on Applied Electro- chemistry in the Massachusetts Institute of Technology. There has been no work in English covering this whole field, and students had either to rely on notes or refer to the sources from which this book is compiled. Neither of these methods of study is satisfactory, for notes cannot be well taken in a subject where illustrations are as important as they are here ; and in going to the original sources too much time is required to sift out the essential part. It is believed that, by collecting in a single volume the material that would be comprised in a course aiming to give an account of the most important elec- trochemical industries, as well as the principal applications of electrochemistry in the laboratory, it will be possible to teach the subject much more satisfactorily. The plan adopted in this book has been to discuss each subject from the theoretical and from the technical point of view separately. In the theoretical part a knowledge of theo- retical chemistry is assumed. Full references to the original sources have been made, so that every statement can be easily verified. It is thought that this will make this volume useful also as a reference book. An appendix has been added, containing the more important constants that are needed in electrochemical calculations. Thanks are due to the following individuals and companies for permission to reproduce cuts, or to use the material in the text, or for both: the American Academy of Arts and Sci- ences ; the American Electrochemical Society ; the Carborun- dum Company ; Wilhelm Engelmann ; Ferdinand Enke ; the Electric Storage Battery Company; the Engineering and Mining Journal; the Faraday Society; the Franklin Insti- 387459 VI PREFACE tute ; Charles Griffin and Company ; Gould Storage Battery Company ; Dr. Eugene Haanel ; the Hanson and Van Winkle Company; Mr. Carl Hering; Mr. Walter E. Holland of Thomas A. Edison's Laboratory ; International Acheson Graphite Company; Wilhelm Knapp; Longmans, Green and Company ; Progressive Age Publishing Company ; Dr. E. F. Koeber, Editor of the Metallurgical and Chemical Engineer- ing ; Julius Springer; Spon and Chamberlain; John Wiley and Sons. TABLE OF CONTENTS CHAPTER I PAGES COULOMETERS OR VOLTAMETERS 1-12 1. General Discussion 2. The Silver Goniometer 3. The Copper Goniometer 4. The Water Goniometer 5. The Silver Titration Goniometer. CHAPTER IT ELECTROCHEMICAL ANALYSIS 13-29 1. Nonelectrolytic Methods 2. Electrolytic Methods. CHAPTER III ELECTROPLATING, ELECTROTYPING, AND THE PRODUCTION OF ME- TALLIC OBJECTS 30-42 1. Electroplating: Nickel Plating ; Copper Plating ; Zinc Plat- ing ; Brass Plating ; Silver Plating ; Gold Plating 2. Gal- vanoplasty : Electrotyping ; Copper Tubes, Foil, and Wire. CHAPTER IY ELECTROLYTIC WINNING AND REFINING OF METALS IN AQUEOUS SOLUTIONS 43-67 1. The Winning of Metals: Copper and Zinc 2. The Elec- trolytic Refining of Metals: Copper Refining; Nickel Refining; Silver Refining ; Gold Refining ; Lead Refining ; Zinc Refining. CHAPTER V ELECTROLYTIC REDUCTION AND OXIDATION .... 68-79 1. Reduction 2. Oxidation. CHAPTER VI ELECTROLYSIS OF ALKALI CHLORIDES 80-136 1. Theoretical Discussion : The Chemical Action of Chlorine on Water and Alkali Hydrate; The Electrolysis of Alkali Chloride on vii TABLE OF CONTENTS PAGES Smooth Platinum Electrodes without a Diaphragm; The Electrolysis of Alkali Chlorides with Platinized Platinum Anodes; The Electrol- ysis of Alkali Chlorides on Carbon Anodes; The Maximum Concen- trations of Hypochlorite and the Maximum Current and Energy Yields of Hypochlorite and Chlorate; The Production of Perchlorates ; The Electrolysis of Alkali Chlorides with a Diaphragm; Decomposition Points and Potentials of Alkali Chloride Solutions; Fluorides, Bromides, and Iodides 2. Technical Cells for Hypochlorite, Chlorate, Hydrate, and Chlorine. CHAPTER VH THE ELECTROLYSIS OF WATER 137-141 CHAPTER VHI PRIMARY CELLS . 142-151 CHAPTER IX THE LEAD STORAGE BATTERY 152-172 1. History and Construction 2. Theory of the Lead Storage Battery. CHAPTER X THE EDISON STORAGE BATTERY 173-184 1. General Discussion 2. Theory of the Edison Storage Battery. CHAPTER XI THE ELECTRIC FURNACE 185-201 1. General Discussion 2. Electric Furnace Design. CHAPTER XII PRODUCTS OF THE RESISTANCE AND ARC FURNACE . . . 202-238 1. Calcium Carbide 2. Carborundum 3. Siloxicon 4. Sili- con 5. Graphite 6. Carbon Bisulphide 7. Phosphorus 8. Alundum 9. Aluminum 10. Sodium and Potassium 11. Calcium. TABLE OF CONTENTS IX CHAPTER XIII PAGES THE ELECTROMETALLURGY OF IRON AND STEEL . . . 239-264 1. General Discussion 2. The Electrothermic Reduction of Iron Ores 3. The Electrothermic Refining of Steel. CHAPTER XIV THE FIXATION OF ATMOSPHERIC NITROGEN .... 265-287 1. Introduction 2. Absorption by Calcium Carbide 3. The Oxidation of Nitrogen 4. The Synthesis of Ammonia 5. Con- clusion. CHAPTER XV THE PRODUCTION OF OZONE 288-314 1. General Discussion : The Maximum Concentration; Yield per Coulomb for Negative Point Electrode ; Yield per Coulomb for Posi- tive Point Electrode ; Yield per Kilowatt Hour for Positive and for Negative Points; Theory of Ozone Formation by Silent Discharge; The Siemens Ozonizer 2. The Technical Production of Ozone. APPENDIX 315-321 Atomic Weights Electrochemical Equivalents Numerical Relation between Various Units Legal Electrical Units. NAME INDEX 323-325 SUBJECT INDEX 327-329 LIST OF ABBREVIATIONS Ann. d. Phys Annalen der Physik. Ann. d. Chem. und Pharm. . . Annalen der Chemie und Pharmacie. Ann. d. Chim. et de Physique . Annales de Chimie et de Physique. B. B Berichte der Deutschen Chemischen Gesell- schaft. Berg- und Hiittenm. Ztg. . . Berg- und Hiittenmanische Zeitung. Chem. News Chemical News. Chem. Zeitung Chemiker Zeitung. C. R Comptes Rendus des Seances de 1'Academie des Sciences. Dingler's polyt. J Dingler's Polytechnisches Journal. Electrochem. and Met. Ind. . Electrochemical and Metallurgical In- dustry. Electroch. Ind. Electrochemical Industry. Elektrotech. Z Elektrotechnische Zeitschrift. El. World Electrical World. Eng. and Min. J Engineering and Mining Journal. Gilbert's Ann Gilbert's Annalen. J. f. prakt. Ch Journal fur praktische Chemie. Journ. of the Franklin Inst. . Journal of the Franklin Institute. J. Am. Chem. Soc Journal of the American Chemical Society. Met. and Chem. Eng Metallurgical and Chemical Engineering. Min. Ind Mineral Industry. Phil. Mag Philosophical Magazine. Phil. Trans Philosophical Transactions. Phys. Rev Physical Review. Pogg. Ann Poggendorff's Annalen. Proc. Am. Acad Proceedings of the American Academy of Arts and Sciences. Proc. Am. Phil. Soc. .... Proceedings American Philosophical So- ciety. Proc. Royal Soc. of Edinburgh Proceedings of the Royal Society of Edin. burgh. Proc. Soc. Arts Proceedings of the Society of Arts, Boston. Trans. Am. Electrochem. Soc. . Transactions of American Electrochemical Society. Z. f. anal. Ch Zeitschrift fur analytische Chemie. xi Xii LIST OF ABBREVIATIONS Z. f. angew. Ch Zeitschrift fiir angewandte Chemie. Z. f. anorg. Ch Zeitschrift fiir anorganische Cheraie. Z. f. Berg-, Hiittenm.- und Salinen-Wesen. Zeitschrift fiir das Berg-, Hut- tenmanische- und Salinen-Wesen in preussische Staaten. Z. f. Elektroch Zeitschrift fiir Elektrochemie. Z. f. phys. Ch Zeitschrift fiir physikalische Chemie. APPLIED ELECTROCHEMISTRY APPLIED ELECTROCHEMISTRY CHAPTER I X COULOMETERS 1 OR VOLTAMETERS 1. GENERAL DISCUSSION AN important application of electrolysis is the determination of the amount of electricity passing through a circuit in a given time. According to Faraday's laws, (1) the magnitude of the chemical effects produced in a circuit is proportional to the quantity of electricity that passes through the circuit, and (2) the quantities of the different substances which separate at electrodes throughout the circuit are directly proportional to their equivalent weights. 2 The first statement is true under all conditions, but the second only for the case that a single sub- stance is liberated on any given electrode. If several sub- stances are deposited together on the same electrode, there is, of course, less of each than if only one is deposited. The electrochemical constant, or the quantity of electricity necessary to deposit one equivalent weight of any substance, has been accurately determined by measuring the amount of silver deposited for a known quantity of electricity. The value of this constant generally accepted is 96,540 coulombs, and is accurate to a few hundredths of a per cent. 3 1 This name was proposed by T. W. Richards, Proc. Am. Acad. 37, 415, (1902). 2 Le Blanc, Electrochemistry, English translation, p. 42, (1907). 3 Nernst, Theoretische Chemie, 6th ed., p. 716, (1909) ; Guthe, Bulletin of the Bureau of Standards, 1, 362, (1905). B 1 2-- ' "* ALLIED.. ELECTROCHEMISTRY It is evident from the above that the amount of electricity passing through a circuit can be determined from the amount of chemical change produced at any electrode if this chemical change can be measured. There are three general methods of making this measurement : (1) by weighing the substance de- posited or liberated, (2) by measuring its volume, and (3) by titration. It seems hardly necessary to call attention to the fact that in any coulometer the current can be computed from the quantity of electricity that has passed through the circuit, if the current has been constant and if the time is measured. Current in amperes equals quantity in coulombs divided by time in seconds. The errors of coulometers are those inherent in the measure- ment of weight and volume or in titration, and also those due to imperfections in the coulometer itself. The latter may come from a variety of causes, such as the liberation of other substances than the one assumed, or the loss of the substance after deposition and before weighing. The errors of each coulometer described below will be pointed out. 2. THE SILVER COULOMETER The silver coulometer is the most accurate of all electro- chemical coulometers. It is for this reason that it is used to determine the electrochemical constant. It consists of a plati- num dish cathode, a neutral silver nitrate solution made by dissolving 20 to 40 grams of nitrate in 100 grams of distilled water, and a pure silver anode. By weighing the platinum dish before and after the current has passed, the amount of electricity may be computed from the value of the electro- chemical equivalent of silver given above. To obtain the best results, the anode should be wrapped in filter paper, 1 in order to prevent any silver mechanically detached from the anode from falling into the platinum dish, or contained in a porous cup, which also separates the anode solution from the cathode. The solution from the anode would deposit too much silver on the 1 Richards, Collins, and Heimrod, Proc. Am. Acad. 35, 143, (1899). COULOMETERS OR VOLTAMETERS cathode, due to the formation of a complex silver ion, prob- ably Ag + , which does not break up at once to the normal ion Ag + and 2 Ag, and which, if deposited, would give too great a quantity of silver. 2 This is the main source of error, and when it is excluded, the mean error of one determination is about 0.03 per cent, for a deposit weighing not less than half a gram. 3 The cou- lometer used by Richards, Collins, and Heimrod is shown in Figure 1. The solution of silver nitrate may be used until a deposit corresponding to 3 grams of silver from 100 cubic centimeters of solu- tion has been reached. The current density must not exceed 0.2 ampere per square centimeter on the anode, or 0.02 ampere per square centimeter on the cathode. The silver ni- trate solution must be thoroughly washed out before weighing, until the wash water gives no test for silver with hydro- chloric acid. The dish is then dried and weighed. The silver deposit from the nitrate solution is crystalline, and does not form a smooth coating, and for this reason there is danger of losing some of the crystals in washing. Silver can be deposited with a smooth surface from the double cyanide of silver and potassium, and it 2 Richards and Heimrod, Proc. Am. Acad. 37, 415, (1902). 3 Ostwald-Luther, Hand- und Hiilfsbuch, 3d ed. p. 497, (1910). FIG. 1. Porous cup coulometer ( actual size) A, glass hook for supporting anode. , glass ring for supporting porous cup. O, silver anode. D, porous cup. E, platinum cathode. APPLIED ELECTROCHEMISTRY has been found that a coulometer using this liquid, on exclud- ing oxygen, gives accurate results without the danger of de- taching any silver in weighing. 4 3. THE COPPER COULOMETER The copper coulometer consists usually of two sheets of copper for anodes, with a thin copper sheet hung between them as cathode, in an acid solution of copper sulphate. It is not so accurate as the silver coulometer for several reasons. In the first place, only 0.29 gram copper is deposited to every gram of silver. This reduces the percentage accuracy of the weight to about one third of the value it would have for an equivalent amount of silver. More important than this are the chemical reactions that tend to change the weight of copper deposited on the cathode from the correct weight. The copper cathode dissolves slightly in acid cupric sulphate, forming cuprous sulphate : Cu + Cu ++ = 2 Cu + , thereby reducing the weight of the cathode. This takes place to a less extent if oxygen is excluded. On the other hand, in a neutral solution the plate gains in weight, due to a covering of cuprous oxide coming from hydrolysis of the cuprous sulphate. With increasing temperature not only does the velocity of the above reaction increase, but also the amount of cuprous ions in equilibrium with cupric ions, and conse- quently more cuprous ions are deposited. Wherever cuprous ions are deposited, the weight of copper is too great, as the electrochemical equivalent of cuprous copper is double that of cupric. The solution generally used in the copper coulometer is that recommended by Oettel, 1 consisting of 1000 grams of water, 150 grams of crystallized copper sulphate, 50 grams of concen- * Farup, Z. f. Elektroch. 8, 669, (1902). 1 Chem. Zeitung, 17, 643, and 677. COULOMETERS OR VOLTAMETERS FIG. 2. Copper coulometer 6 APPLIED ELECTROCHEMISTRY trated sulphuric acid, and 50 grams of alcohol. The alcohol drives back the dissociation of the cupric sulphate, reducing the concentration of the cupric ions and therefore of the cu- prous ions in equilibrium with them. 2 For ordinary purposes the exclusion of air is not necessary. The current density on the cathode should lie between 2 and 20 milliamperes per square centimeter. The advantages of the copper over the silver coulometer are its greater cheapness and the greater adhesiveness of the deposit on the cathode. The average error of a single determination is from 0.1 to 0.3 per cent. 3 A convenient form of the copper coulometer is shown in Fig- ure 2. The inside dimensions of the glass vessel are approxi- mately 4.3 centimeters in width, 16 centimeters in height, and 17 centimeters in length. 4. THE WATER COULOMETER The water coulometer measures the quantity of electricity passing through a circuit by the amount of water decomposed between unattackable electrodes dipping in a solution through which the current flows. The amount of water decomposed may be determined by measuring the loss in weight of the coulometer, by measuring the total volume of gas produced, or by measuring the volume of either one of the gases separately. The decomposition of water by the electric current was first observed by Nicholson and Carlisle l in 1800. In 1854 Bunsen 2 used a water coulometer in which the loss in weight was deter- mined ; and since then others have devised coulometers on the same principle. 3 Figure 3 shows a convenient form of the apparatus, having a drying tube sealed directly to it ; for be- fore leaving the cell the gases must, of course, be thoroughly 2 Foerster and Seidel, Z. f. anorg. Ch. 14, 135, (1807). 8 Ostwald-Luther, Hand- und Hiilfsbuch, 3d ed. 497, (1910). 1 Gilbert's Ann. 6, 340, (1800). 2 Pogg. Ann. 91, 620, (1854). 8 L. N. Ledingham, Chein. News, 49, 85, (1884). COULOMETEKS OR VOLTAMETERS Glass wool Sealed joint dried so that no water vapor is carried off with them. It is evident that this instrument cannot give great accuracy on account of the relatively small change in weight produced by the passage of an amount of electricity equal to the electrochemical constant. In the case of water the change in weight is only 9 grams, as com- pared with 31.2 grams of copper and 107.9 grams of silver. The errors inherent in the instrument itself are due to the formation of other products than hydrogen and oxygen. If a solution of sulphuric acid is used between platinum elec- trodes, the oxygen liberated on the anode contains a certain amount of ozone. 4 Persulphuric acid, H 2 S 2 O 8 , and hydrogen peroxide, due to the oxidation of water by the persul- phuric acid, are also produced. The production of persulphuric acid is a maximum when the concentration of the solution is between 30 and 50 grams of sulphuric acid to 100 grams of water. 5 For this reason a 10 to 20 per cent solution of sodium hydrate is often used, in which none of the above disturbing reactions occur. The presence of even a small amount of salt of a metal with two different valences, such as iron, may cause a very large error. Table 1 shows what the magnitude of this error is for iron impurities. 6 4 Schonbein, Pogg. Ann. 50, 616, (1840). 6 Franz Richarz, Ann. d. Phys. 24, 183, (18oo); 31, 912, (1887). e Elbs, Z. f. Elektroch. 7, 261 (1900). FIG. 3. Water coulometer 8 APPLIED ELECTROCHEMISTRY TABLE 1 IRON CONTENT IN PER CENT CURRENT DENSITY PER SQUARE DM. Loss IN DETONATING GAS IN PER CENT 1.0 2.23 48.3 1.0 0.228 97.4 0.1 6.4 3.0 0.1 0.35 25.1 0.01 2.22 1.7 0.360 6.3 It is to be noticed that this error is diminished by increasing the current density. Sulphuric acid of 1.14 specific gravity has been shown by F. Kohlrausch 7 to give results as accurate as the measurements themselves in coulometers where the total volume of gas is measured. He simultane- ously devised a form of coulom- eter shown in Figure 4. The glass tube is 4 centimeters in diameter and is divided into units of 5 cubic centimeters. The base contains 500 cubic centimeters. The anode is platinum foil, 4 centimeters long and 1.7 centimeters wide, placed between two cathodes of the same size. To refill the tube it is simply turned upside down. A thermometer is sealed in for determining the tempera- ture of the gas. On account of the limited volume of this ap- FIO. 4. Kohlrausch water couiometer paratus, large quantities of elec- 7 Elektrotech. Z. 6, 190, (1885). COULOMETERS OR VOLTAMETERS 9 tricity cannot be measured; it is intended for the measurement of currents between 3 and 30 amperes. The relation between the volume of gas generated in one second, saturated with water vapor at the vapor pressure corresponding to a sulphuric acid solution of specific gravity 1.14, and the current is as follows: For 20 and a pressure of 72.5 centimeters of mercury, one ampere in one second produces 0.2 cubic centimeter of gas, including the water vapor. Therefore, under these conditions of temperature and pressure, the number of cubic centimeters of gas generated per second, when multiplied by 5, gives the current in amperes. The corrections for the volume in thou- sandths of a cubic centimeter for different temperatures and pressure are given in Table 2. TABLE 2 Corrections, in Thousandths of a Cubic Centimeter, for Reducing the Volume of Gas generated in One Second to the Value which, multiplied hy 5, gives the Current. Specific Gravity of Sulphuric Acid : 1.14 TEMP. CENTIGRADE DEGREES 700 mm. 710 mm. 720 mm. 730 mm. 740 mm. 750 mm. 760 mm. 10 9 24 38 53 68 82 97 11 5 19 33 48 63 78 93 12 1 15 29 44 59 73 88 13 -4 10 24 39 54 69 83 14 -8 6 20 35 49 64 78 15 -13 2 16 30 44 59 73 16 -17 -3 11 26 40 54 68 17 -22 7 7 21 35 49 63 18 -26 -12 2 16 30 45 59 19 -31 -17 -3 11 26 40 54 20 -35 -21 -7 7 21 35 49 21 -40 -26 -12 2 16 30 44 22 -44 -30 -17 -3 11 25 39 23 -49 -35 -22 -8 6 20 34 24 -54 -40 -26 -12 1 15 29 25 -58 -45 -31 -17 -4 10 24 10 APPLIED ELECTROCHEMISTRY The following example will illustrate the use of this table. Barometer, 754 millimeters of mercury. Height of sulphuric acid in tube 112 millimeters of mercury. Pressure in gas = 754 - -^ = 745. Temperature of gas : 17. 8. Volume of gas 198.0 cubic centimeters. Correction : + 0.038 x 198.0 = 7.5 cubic centimeters. 205.5 cubic centimeters. Duration of experiment : 39 seconds. Therefore in one second 5.27 c.c. of gas were generated. Current = 5.27 x 5 = 26.3 amperes. > k ._. On comparison with a tangent galvanom- X^A eter the current indicated YJU by this coulometer was found on an average to be \ per cent low. In order to avoid correc- tion for the height of the solution, the instrument may be made like a Hem- pel gas analysis burette, as shown in Figure 5. A very convenient form of water coulometer has been devised by F. C. G. Miiller, 8 shown in Figure 6. The whole apparatus is placed in a water bath, so that the temperature of the gas can be determined. A is the electrolytic cell filled with barium hydrate, which does not foam like sodium or potassium hy- drate. Fio. 5. Water cou- lometer ceiver. F is the gas re- ~~ mi ,, FiG.6. Miiller's water The three-way coulometer 8 Z. f. d. phys. und chem. Unterricht, 14, 140, (1901). COULOMETERS OR VOLTAMETERS 11 stopcock at the top allows the gas to escape through H when no measurement is to be made. By turning the stopcock at a given second, the gas passes into H, which is previously filled with water to the upper mark. When H is filled with gas, the stopcock is turned to allow the gas to pass out H and the time noted. This apparatus can thus be left connected in the circuit and a measurement made at any time. The water coulometer may be transformed into a direct read- ing ammeter by a method first applied in 1868 by F. Guthrie. 9 If the gas is allowed to escape through a small hole, a definite pressure in the instrument is developed, depending on the cur- rent and size of the hole. The pressure is measured by a mercury or water manometer. This same principle has been rediscovered by J. Joly, 10 Bredig and Hahn, 11 and Job. 12 In Bredig and Hahn's apparatus the gas escapes through capillary tubes, and by using a tube with different bores the range of the instrument is varied. Their apparatus is accurate to about 5 per cent. 5. THE SILVER TITRATION COULOMETER The silver titration coulometer of Kistiakowsky l is some- times convenient where the current does not exceed 0.2 ampere and where the duration of the experiment does not exceed an hour. A silver anode is dissolved in a 10 per cent potassium nitrate solution by the passage of the current, and is then titrated. In the improved form the silver anode is at the bottom of a tube 18 to 22 centimeters long, 3.5 centimeters in diameter at the top, and 1 centimeter at the bottom. The cathode is of copper and dips in a 7 per cent copper nitrate solution to which ^ of its volume of a 10 per cent potassium 9 Phil. Mag. 35, 334, (1868). 10 Proc. Royal Dublin Soc. 7, 559, (1892). 11 Z. f. Elektroch. 7, 259, (1901). 12 Z. f. Elektroch. 7, 421, (1901). i Z. f. Elektroch. 12, 713, (1906). 12 APPLIED ELECTROCHEMISTRY nitrate solution has been added. This solution is contained in a porous cup at the top of the tube. After the experiment the potassium nitrate solution containing the dissolved sil- ver is drawn off and titrated with 0.02 normal potassium thiosulphate, and a saturated iron alum solution as indicator. The error of a single determination may amount to 0.5 per cent. In the original form, which is the one still generally used, the cathode is of platinum and dips into a J to ^ normal solution of nitric acid. The division between the acid and the nitrate is shown in Figure 7 by the dotted line. In order to have the silver dissolve with 100 per cent efficiency, it should be freshly deposited electrolytically ; 2 also, all of the anode should be the same distance from the cathode, as shown in the figure ; otherwise the current density will be too great on the part nearest the cathode, and bubbles of gas may be given off. It is conven- ient to have the anode made of a platinum spiral of the form shown, on which a little more silver is deposited electrolytically before a measure- be dissolved off in the measure- tion coulometer ment. F ^S: ment than 2 Ostwald-Luther, Hand- und Hiilfsbuch, 3d ed. p. 600, (1910). CHAPTER II ELECTROCHEMICAL ANALYSIS 1. NONELECTROLYTIC METHODS THERE are four different electrical methods of quantitative analysis. These are (1) potential measurements, which give a means of determining the concentrations of ions too dilute to determine gravimetrically ; (2) conductivity measurements, which is a method very convenient for determining concentra- tions of solutions; (3) titration with a galvanometer in place of an ordinary indicator, and finally (4) the ordinary electro- analysis, in which the metal is deposited on a platinum electrode and weighed. The principle of the first method, originally pointed out by Ostwald, 1 is as follows : Suppose the concentration of silver chloride in its saturated solution is desired. If the electro- motive force of the cell Ag | T V NAgN0 3 1 T V NKN0 3 1 saturated solution of AgCl | Ag were measured, and the concentration ^ of the silver ions in the nitrate were known, as it is from conductivity measure- ments, the concentration c z of the silver chloride ions could be computed by the Nernst formula In practice some conducting salt is added to the silver chloride solution in order to lower the resistance of the cell. If potas- i Lehrbuch, 2d ed. II, 879. 13 14 APPLIED ELECTROCHEMISTRY sium chloride is chosen, the solubility of silver chloride is reduced, but its value in pure water can be computed from this result. 2 If potassium nitrate were used, no reduction in the solubility would take place. Where the concentration of the salt is so small, the ion concentration is very nearly equal to the total concentration on account of the fact that the salt is nearly 100 per cent dissociated. Other instances where this method of measuring ion concentrations has been found use- ful are in the determination of the solubility of mercurous chloride from the electromotive force of the cell : Hg | Hg.CI, in T V NKC1 1 T V NHg 2 N 2 O 6 + HNO 3 1 Hg, and from this result the solubility of mercurous sulphate from the electromotive force of the cell, 3 ""^ 'Hg. and NKNO and NKNO These examples are sufficient to illustrate the method. Some of the errors that attend these measurements may now be men- tioned. One difficult}'' is to get different electrodes of the same metal to show exactly the same electromotive force when placed in the same solution of one of their salts. This seems to depend on the surface of the metal, and some method has to be used to make them as nearly identical as possible. This can often be accomplished by using an electrode covered electrolyti- cally with a layer of the metal, or if the metal is more electro- positive than mercury, amalgams of equal concentrations may be used. 4 The electrolytic solution pressure is thereby some- what changed, but by the same amount for each electrode, and since the electrolytic solution pressure drops out, the resulting electromotive force is unaffected. Another method of obtaining constant results is to use the metal in a finely divided form. This may be done by depositing electrolytically with a high- current density or by decomposing some compound of the metal a Goodwin, Z. f. phys. Ch. 13, 641, (1894). Wilsmore, Z. f. phys. Ch. 35, 20, (1900). 4 Goodwin, I.e. p. 676. ELECTROCHEMICAL ANALYSIS 15 in question. 5 Another source of error is the potential at the junction of the different solutions, but this can generally be either calculated or reduced to an insignificant amount by adding some indifferent salt 6 or by connecting the liquids with saturated solutions of potassium chloride 7 or ammonium nitrate. 8 A method based on potential measurement has been worked out for determining the amount of carbonic acid in gases. 9 The gas bubbles through a solution of bicarbonate, and the result- ing hydrogen ion concentration of the solution is determined by potential measurements, from which the partial pressure of the carbonic acid can be computed. The principle involved in determining the amount of sub- stance in a solution by conductivity measurement 10 is the same as when any other physical property, such as specific gravity, is used for the purpose ; that is, the relation between the conductivity and quantity of substance in solution must be known. These data have already been obtained in a large number of cases and have been collected by Kohlrausch and Hol- born. If the solution contains a single electrolyte whose con- ductivity at given concentrations has already been determined, all that is necessary is to interpolate graphically or arithmetically in the table. If, however, there is a maximum conductivity, as in the case of sulphuric acid, there would be two possible con- centrations for a given value of the conductivity. It is easy to tell on which side of the maximum such a solution lies by diluting a little and redetermining the conductivity. If the solution were more dilute than corresponds to the maximum value, further dilution would decrease the conductivity ; if less dilute, the conductivity would be increased. In case the solu- tion has a concentration near that of maximum conductivity, 5 Richards and Lewis, Z. f. phys. Ch. 28, 1, (1899) ; also Lewis, J. Am. Chem. Soc. 28, 158, (1905). 6 Bugarszky, Z. f. anorg. Ch. 14, 150, (1897). 7 Bjerrum, Z. f. phys. Ch. 53, 428, (1905). 8 Gumming, Z. f. Elektroch. 13, 17, (1907). 9 Bodlander, Jahrb. d. Elektroch. 11, 499, (1904). 10 See Kohlrausch and Holborn, Das Leitvermogen der Elektrolyte, p. 124, (1898). 16 APPLIED ELECTROCHEMISTRY where the determination would be inaccurate, it can be diluted enough to remove it from this point, and the contents of the new solution determined. From this the concentration in the original one can be calculated. This method has been shown to be useful in the determina- tion of impurities in sugar and of mineral waters. 11 On account of the fact that the equivalent weights of the impurities likely to be present in mineral waters vary only within certain limits, it has been found that the quantity of the impurities can be estimated with a fair degree of accuracy from conductivity without analyzing the water to see which of the usual impurities are present. This method is also useful in the case of mixtures of two salts when the conductivity of the mixture is the arithmetical mean of the single conductivities. This is often the case with nearly related compounds, which are generally difficult to sepa- rate chemically. For two substances for which this rule holds, having at equal concentrations the specific conductivities K and jfiQ, the conductivity of a mixture of the same total concen- tration would have the conductivitv K = K \P\ + K iP*. By Pi+Pt this means it has been found possible to analyze satisfactorily mixtures of potassium chloride and bromide, and sulphates of potassium and rubidium. 12 Conductivity has also been applied extensively for the determination of the solubility of very insoluble salts. 10 The use of a galvanometer as an indicator depends for the end point either on a sharp change in the resistance of the cell containing the solution titrated or in the change in the electro- motive force on electrodes dipping in this solution. An example of the first case is the titration of silver nitrate with a standard solution of potassium chloride. 13 A measured quantity of a standard solution of potassium chloride is placed in a beaker with two silver electrodes. In series with the two electrodes u Reichert, Z. f. anal. Ch. 28, 1, (1889). 12 Erdmann, B. B. 30, 1175, (1897). 18 Salomon, Z. f. Elektroch. 4, 71, (1898). ELECTROCHEMICAL ANALYSIS 17 are connected a galvanometer and a source of electromotive force, which must be less than the decomposition value of the potassium chloride. On closing the circuit, only a very small residual current will be detected. On adding a little of the silver nitrate to the solution, silver chloride is precipitated, and a certain amount of silver ions, corresponding to the solubility of the chloride, will be in solution. We now have the cell Ag | AgCl solution | Ag, which has no decomposition point, but the quantity of silver is so small that the large resistance prevents the current from increasing to any great extent. As nitrate is added, the quantity of silver in solution changes very little until the last of the potassium chloride is used up. The first drop of silver nitrate in excess now increases the silver ions enormously, and there is a corresponding large increase in current, due to the reduced resistance of the cell. The following table shows the sharpness of the change : 13 CUBIC CENTIMETERS OF AoNO 8 GALVANOMETER BEADING 3.00 15 4.40 20 5.00 16 5.50 21 5.60 20 5.65 42 The use of a galvanometer as indicator when the electro- motive force changes suddenly at the end point is illustrated by the following examples : 14 Suppose two beakers, one containing a tenth normal solution of mercurous nitrate, the other a definite quantity of mercurous nitrate solution to be titrated, are con- nected by a siphon containing tenth normal potassium nitrate. The bottom of each beaker is covered with a layer of mercury which makes contact with a platinum wire sealed in the glass. w Behrend, Z. f. phys. Ch. 11, 482, (1893). 18 APPLIED ELECTROCHEMISTRY Such a cell would have the electromotive force R T log -l, where '2 e l is the concentration of the mercury ions in the tenth normal solution and c z is their concentration in the unknown solution. If tfj is equal to c 2 , the electromotive force would be zero, but in general e l and OC1- + H 2 O, (4) for which the equilibrium constant is C OH _ This is the hydrolysis constant of the hypochlorite. The value of K 3 can be obtained from the dissociation constant of hypo- chlorous acid : 6 r r K = ^oci- ^H+ = 3. 7 x lQ-8, 2 Luther, Z. f. Elektroch. 8, 602, (1902). 3 Foerster, Elektrochemie wasseriger Losungen, p. 341, (1905). 4 Foerster and Miiller, Z. f. Elektroch. 8, 921, (1902). The value of K w is taken from van Laar, Theoretische Elektrochemie, p. 174, (1907). 6 Sand, Z. f. phys. Ch. 48, 610, (1904). 82 APPLIED ELECTROCHEMISTRY and the dissociation constant of water: K W = C H+ COH- = 1.4 X 10-". Dividing K w by K 4 , KW _ CH+ COH- CHOCI _ _ 1.4 X 10~ 14 _. 3 g x K 4 C OC1 -.C H+ "8.7x10-8 When therefore any quantity of chlorine acts on alkali hydrate, the resulting quantities of hydrate, chlorine, chloride, hypo- chlorous acid, and hypochlorite are determined by the equilibria represented by equations (2) and (4). Only when there are at least two equivalents of hydrate to one mole of chlorine does the following reaction hold : C1 2 + 2 NaOH = NaCl + NaOCl + H 2 O. (5) This is the sum of equations (2) and (4), and is the one usually given to represent the reaction between chlorine and hydrate. Since the equilibria represented by equation (3) and the equa- tion for the value of K 3 exist simultaneously, the values of C OII _ and CHOCI are the same in both. From the equation for K 3 , CHOCI _ vr C OC1 _ P 3 * 77 2 ^OH- *-* OH- and combining this with the equation for K 2 , ^Cl- ^OH- ^ OH- This equation is convenient for predicting what effect a change in the concentration of one substance will have on that of the others. From equation (6) it would seem that for a given value of C cla and C C1 _, the value of C OC1 _ could be increased in propor- tion to the value of C OH _. This would be true, if the concen- tration of the hypochlorite, C OC1 _, were not limited by another reaction, the oxidation of hypochlorite to chlorate by hypo- chlorous acid, according to the equation : 6 2 HOC1 + QC1- = ClOg + 2 Cl- + 2 H+. (7) Foerster and Jorre, J. f . prakt. Ch. 59, 53, (1899) ; Foerster, ibid. 63, 141, (1901). ELECTROLYSIS OF ALKALI CHLORIDES 83 The free hydrochloric acid then sets free an equivalent amount of hypochlorous acid according to the equation : 2 H + + 2 Cl- + 2 OC1- = 2 HOC1 + 2 Cl~, (8) and the hypochlorous acid thus set free oxidizes more hypo- chlorite. This process continues until all of the hypochlorite has been changed to chlorate. Substituting the numerical values of K 2 and K 3 in (6), we have 1.4 xlO-tto^- (9) In order to illustrate the use of the above equation, the rela- tive concentrations of chlorine ions, hydroxyl ions, free chlorine, and hypochlorous acid in a neutral solution normal with respect to hypochlorite ions will be calculated. From the value of K 3 , the values of C HOC1 and C OH _ are each 6.2 10~ 4 mole per liter, and from (9) the value of the fraction ^- is 3.6 x lO' 11 . If ^ci- the concentration of chlorine ions is also normal, that of the free chlorine is only 3.6 x 10~ n mole per liter. 4 If chlorine is led into a solution of alkali hydrate, nothing but hypochlorite and chloride are produced as long as some of the hydrate remains unneutralized. This is because the excess of hydroxyl ions drives back the hydrolysis of the hypochlorite and therefore prevents the formation of a free hypochlorous acid. When an amount of chlorine equivalent to the hydrate has been added, there is still so small a quantity of free hypo- chlorous acid present that the solution is fairly stable. An excess of chlorine, however, increases the concentration of the free hypochlorous acid to such an extent that the hypochlorite is rapidly oxidized to chlorate, according to equation (7). The fact that an excess of chlorine was necessary to produce chlo- rate was first discovered by Gay-Lussac. 7 The addition of a small quantity of free acid would have the same effect as an excess of chlorine, for it would set free hypochlorous acid. 7 Liebig Ann. 43, 153, (1842). 84 APPLIED ELECTROCHEMISTRY If chlorate were formed only by means of free hypochlorous acid, hypochlorite would be more stable the greater the excess of hydroxyl ions in the solution. Chlorate is produced, how- ever, slowly in alkaline solutions, presumably by the reaction 3 NaOCl = NaClO 3 + 2 NaCl. (10) Hypochlorous acid breaks up in exactly the same way, when it decomposes of itself. The solution has to be heated to 70 C. to make this reaction proceed with an appreciable velocity, 6 and it is also catalyzed by light. With increasing alkalinity the velocity of the reaction increases somewhat, and it is always accompanied by the reaction : 2 NaOCl = O 2 + 2 NaCl. (11) The last reaction is catalyzed by some metallic oxides, espe- cially by the oxide of cobalt, to such an extent that all of the hypochlorite can be decomposed in this way without forming any chlorate. 6 Perchlorate cannot be formed by the further action of chlorine on chlorate, but is produced by the decomposition of chlorate, as will be explained below. The Electrolysis of Alkali Chloride on Smooth Platinum Elec- trodes without a Diaphragm If a concentrated neutral solution of alkali chloride is electro- lyzed between smooth platinum electrodes, the alkali is de- posited on the cathode and reacts with the water according to the equation : 2 Na + 2 H 2 O = 2 NaOH + H 2 . (12) The hydrogen produced escapes, unless it is used up in reduc- ing some substance in the solution. On the anode, chlorine is liberated from the ionic form to free chlorine, as follows : 2C1- + 2F=C1 2 . (13) The liberated chlorine partly dissolves in the water and at first partially escapes from the solution. Soon, however, the alkali hydrate produced at the cathode and the dissolved ELECTROLYSIS OF ALKALI CHLORIDES 85 chlorine are brought together by the stirring produced by the escaping hydrogen, and after this no more chlorine escapes from the solution. Chlorine and alkali hydrate are produced in equivalent quantities, so that the equation (5), C1 2 + 2 NaOH = NaCl + NaOCl 4- H 2 O, is practically quantitative. It is evident that only 50 per cent of the chlorine liberated is obtained in the active form as hypo- chlorite. As the electrolysis proceeds, the hypochlorite becomes more and more concentrated, until finally a limiting concentra- tion is reached, whose value is determined by a number of factors, such as the material of the anode, the current densities on the anode and cathode, the temperature, and the original concentra- tion of the chloride solution. This is due to the fact that the hypochlorite, almost from the start, is also decomposed by the current, and this decomposition increases as the concentration of the hypochlorite increases, until the amount decomposed is just equal to the amount produced. This decomposition takes place in two ways ; at the cathode the hypochlorite is reduced by the hydrogen as follows : NaOCl + H 2 = H 2 4- NaCl, (15) and at the anode the hypochlorite ion is liberated, since it is more easily discharged than the chlorine ion, 1 and reacts with the water, producing chlorate and oxygen according to the following reaction : 2 6 CIO- + 3 H 2 = 2 C10 3 - + 4 Cl- + 6 H+ + lO a . (1Q) This has been called the anode chlorate formation, since it takes place only on the anode and not throughout the solution. It may help in understanding the chloride electrolysis if, before discussing it further, a method of analysis is explained which has been extensively used in the study of this subject for following the reactions taking place during the electrolysis. 1 Foerster and Miiller, Z. f. Elektroch. 8, 634, (1902). 2 Foerster and Miiller, Z. f. Elektroch. 8, 667, (1902). This equation is con- sidered fairly well established, as will be shown below, though other explanations of the results are possible. 86 APPLIED ELECTROCHEMISTRY This consists in analyzing the gas evolved from the cell and comparing the amount of hydrogen and oxygen in it with that evolved by the same current from a water coulometer. 3 If there is less hydrogen from the chloride cell than from the coulometer, the difference must have been used in reducing the hypochlorite, according to equation (15), as this is the only reducible substance in solution. The oxygen in the gas evolved from the cell con- taining the chloride solution must be due to the discharge of the hypochlorite ion, which reacts with the water according to equa- tion (16), producing chlorate and oxygen. Oettel believed that the reaction was simply the evolution of oxygen according to the equation : 2 CIO- + H 2 = 2 HOC1 + i 2 , (IT) and he therefore called this portion of the current loss " water decomposition," but this view has since been found to be incor- rect. Since the proportion of oxygen evolved to the hypo- chlorite ions discharged is the same in either case, Oettel's calculations will not be changed, but the explanation of the oxygen evolution will be given by equation (16) in place of (17). According to equation (17), the oxygen evolved is pro- portional simply to a current loss without destroying hypochlo- rite already formed, while according to (16) it is proportional to a fraction of the current that changes hypochlorite to chlorate. The following example, illustrating the use of gas analysis for determining the yield in hypochlorite as the electrolysis pro- gr,esses, is taken from Oettel. 3 The cell containing the chloride solution was connected in series with a water coulometer. During a given time, at the beginning of the electrolysis, 60 cubic centimeters of gas were evolved from the coulometer and 32 cubic centimeters from the chloride solution. In the coulometer, 40 cubic centimeters of the gas must have been hydrogen. By analysis it was found that the gas from the chloride solution had the following com- position : 30 cubic centimeters of hydrogen, 1.6 of oxygen, and 0.4 of chlorine. This shows a difference in the amount of hy- * I\ Oettel, Z. f. Elektroch. 1, 354, (1894). ELECTROLYSIS OF ALKALI CHLORIDES 87 drogen in the two cells of 10 cubic centimeters. This amount must therefore have been used to reduce the hypochlorite already formed. Since 40 cubic centimeters of hydrogen repre- sents the total current, or 100 per cent, the loss of current due to reduction was 100 x |$ = 25 per cent. The loss due to the evolution of chlorine equals 100 x - 1 , or 1 per cent. The 1.6 cubic centimeters of oxygen are equivalent to twice as much hydrogen, or 3.2 cubic centimeters. The loss of current by o n changing hvpochlorite to chlorate was therefore 100 x = 8 40 per cent. The current used to produce hypochlorite is propor- tional to the amount of hydrogen evolved from the chloride solution, diminished by the quantity of chlorine evolved, and twice the amount of oxygen: 30 (3.2 -j- 0.4)= 26.4 cubic centimeters. The current yield is therefore 100 x J| = 66 per cent. This, of course, means that 66 per cent of the current produces hypochlorite according to equation (5) : Cl a + 2 NaOH = NaCl + NaOCl + H 2 O. The rest of the current destroys hypochlorite already produced, or produces chlorine which escapes from the cell. Chlorine is evolved, however, only at the very beginning of the electrolysis, before the hydrate and chlorine have had time to mix. The following table sums up the results of this calculation : Current used to produce hypochlorite 66 per cent Current used to reduce hypochlorite 25 per cant Current loss by changing hypochlorite to chlorate .... 8 per cent Current loss due to evolution of chlorine 1 per cent 100 per cent The curves in Figure 24 4 will illustrate the results of the elec- trolysis of a neutral 4.37 normal sodium chloride solution with a current density on the anode of 0.075 ampere per square centi- meter and on the cathode of 0.18 ampere per square centimeter. The electrolysis was continued for 18 hours, but the plots are given for only 8 hours, as no change in the direction of the * Miiller, Z. f. anorg. Ch. 22, 33, (1900), and Z. f. Elektroch. 6, 14, (1899). 88 APPLIED ELECTKOCHEMISTRY FIG. 24. Electrolysis of a neutral, 4.37 normal sodium chloride solution curves took place during the following 10 hours. The quanti- ties of hypochlorite and chlorate were determined by direct analysis, and are plotted in terms of oxygen contained by each in grams per liter. The corresponding scale of ordi- nates is on the left. The other curves were obtained by gas analysis as described above. The scale of ordiiiates for these is given on the right, in per cent. It will be seen that the fraction of the current used in evolving oxygen and for re- duction, and the concentration of the hypochlorite become constant at the same time. At first the concentration of the chlorate remains low, but increases steadily as soon as the concentration of the hypo- chlorite becomes constant. This shows that hypochlorite is the first product of the electrolysis and that it is the starting point for the formation of chlorate ; also that it is responsible for the evolution of oxygen, as would be expected from equa- tion (16). The same general effect is produced by electrolysis at 50 C., except that the concentration of the hypochlorite becomes con- stant at a lower value. This is due to the increase in the hydrolysis of the chlorine as the temperature rises, thus pro- ducing a greater concentration of hypochlorite ions on the anode from the beginning. The quantity of hypochlorite ions that has to be supplied to the anode from the solution before they are discharged is therefore less than at a lower temper- ature; consequently the concentration in the solution will not reach as high a value as in the cold solution before the amount of hypochlorite decomposed equals the amount produced. 6 6 Foerster, Elektrochemie wasseriger Losungen, p. 364, (1905). ELECTROLYSIS OF ALKALI CHLORIDES 89 Both the reduction of the hypochlorite at the cathode and the discharge of the hypochlorite ion on the anode are made more difficult by increasing the current density, as will be seen from the following considerations. The greater the quantity of chlorine coming from the anode, the more it tends to prevent the hypochlorite from reaching the anode, where it would be discharged, 6 and the smaller the cathode is made, the less oppor- tunity will the hypochlorite have of coming in contact with nascent hydrogen. This is the explanation of the experimen- tal fact that increasing the current density on the cathode low- ers the reduction, and on the anode it makes the evolution of oxygen less in the first stages of the electrolysis, which is equiv- alent to making the concentration of hypochlorite attainable greater. In a dilute solution of chloride, the maximum hypochlorite concentration is less than in a concentrated solution, because at a given concentration of chloride the hypochlorite must carry relatively more of the current than when there is a greater amount of chloride present, and this results in its being changed to chlorate. Table 10 illustrates the effects of temperature, current density, and concentration changes on the electrolysis of alkali chloride solutions. The reduction of the hypochlorite can be nearly entirely pre- vented by the addition of a small amount of potassium chro- mate to the solution. 6 Under the action of the current a thin diaphragm is produced that gives the cathode a brownish yel- low appearance when compared with a fresh piece of platinum, and which gives a test for chromium when dissolved in nitric acid. 7 This diaphragm is probably an oxide of chromium, since a cathode of metallic chromium does not prevent reduc- tion. Potassium chromate is as effective with a low-current density as with a high density. The curves 8 in Figure 25 show the effect of adding 0.18 per cent of chromate to a solution con- 6 E. Miiller, Z. f. Elektroch. 5, 469, (1899); Imhoff, German Patent, 110,420, (1898). 7 E. Mtiller, Z. f. Elektroch. 7, 401, (1901). 8 E. Miiller, Z. f. Elektroch. 5, 470, (1899). 90 APPLIED ELECTROCHEMISTRY taining 30 per cent of sodium chloride. The broken lines refer to the solution without the chromate. The current density on the anode in both cases was 0.075 ampere per square centi- meter ; on the cathode, 0.18 ampere. The temperature was from 42 to 50 C. 11X) 80 CO i 40 20 C \ V s*^ CURRENT YIELD r REDUCTION + .+.- N N^ CURRENT YIELD -+ > | , OXYGEN EVOLUTION F H> s^ P - OXYGEN EVOLUTION ---- 1- REDUCTION 1 1 4 8 12 16 20 24 FIG. 25. Electrolysis of sodium chloride Full lines refer to solutions containing 0.18 per cent chromate, broken lines to solutions containing no chromate When potassium chromate is added, the whole loss in current will therefore be due to oxygen evolved according to equation (16), which may be written : 6 CIO- + 3 H 2 O + 6 F = 6 H + + 2 C1O 8 - + 4 Cl~ + 1J O 2 But 12 equivalents of electricity are required to produce 6 equiv- alents of hypochlorite, according to equation (5), which may be written : 12 Cl- + 12 F + 12 NaOH = 6 NaCl + 6 NaOCl + 6 H 2 O, while 6 equivalents are required to discharge the hypochlorite ions required by (16). If as much hypochlorite is to be decomposed by (16) as is produced by (5), it is evident that twice as much of the current must be used in producing hypochlorite as is used in changing it to chlorate. That is, f of the current produces active ELECTROLYSIS OF ALKALI CHLORIDES 91 oxygen in the solution and -J produces free oxygen, at the same time changing the active oxygen from hypochlorite to chlorate, according to (16). Therefore, excluding reduction and the for- mation of chlorate by equation (7), when the concentration of the hypochlorite has reached a maximum, in a neutral or slightly alkaline solution, 33.3 per cent of the current will be used to produce free oxygen, and 66.7 per cent to produce active oxygen in the solution. 9 The oxygen evolution can never be greater than 33.3 per cent unless the concentration of the solution is small, in which case oxygen would be evolved by the discharge of hydroxyl ions. Except for these points, this relation is inde- pendent of the other conditions of the experiment, such as tem- perature, current density, and, within certain limits, the concentration. This is illustrated by the results in Table 10, 10 column 5. TABLE 10 Solution : 4.8 Normal NaCl and 2 Grams K 2 CrO 4 per Liter LIMITING CONC. OF OXYGEN IN NaClO. GRM. PER PER CENT OF CURRENT PRO- AiMPERES PER 100 c.c. DUCING OXYGEN TEMPERATURE SQ. CM. ON ANODE Smooth Platinized Smooth Platinized Pt. Anode Pt. Anode Pt. Anode Pt. Anode 13 0.017 0.34 0.61 33.3 31.0 13 0.17 0.68 0.89 30 to 33 34.5 50 0.017 0.17 0.31 30 to 31 22.0 50 0.17 0.42 0.64 30 28.0 The Solution Changed to One 1.7 Normal NaCl and Containing 2 Grams K 2 CrO 4 per Liter 13 0.017 0.28 0.48 34 33.5 13 0.17 0.47 0.65 32.5 35 50 0.017 0.15 0.23 33.5 29 to 31 50 0.17 0.35 0.40 33 33 to 34 75 0.017 0.09 0.15 34 25 to 27 75 0.17 0.25 0.32 35 to 36 29 9 Foerster and Miiller, Z. f. Elektroch. 9, 199, (1903). 10 Foerster and Miiller, Z. f. Elektroch. 9, 196, (1903). 92 APPLIED ELECTROCHEMISTRY In the above experiments, when the oxygen evolution is less than 33.3 per cent, hypochlorite is lost by the secondary forma- tion of chlorate. Columns 3 and 5 show that the maximum concentration of hypochlorate is different under different con- ditions, but that when this concentration is reached, the frac- tion of the current used in oxygen evolution is practically the same under widely differing conditions. If the solution of sodium chloride is made acid with hydro- chloric acid at the beginning of the electrolysis, the first effect of electrolysis is to decompose the acid until the solution be- comes nearly neutral. 10 There always remains a small quantity of the free acid throughout the solution, however, liberating free hypochlorous acid, which oxidizes the hypochlorate to chlorate through the entire solution, according to equation (7). This gives a method of increasing th&^ yield in chlorate over that attainable in neutral or alkaline solutions, in which it has been shown above that the maximum yield is 66.7 per cent. If, before the maximum concentration of hypochlorite has been reached, a quantity of acid is added to the solution which is equivalent to only a fraction of the hypochlorite in the solution, the latter is completely oxidized to chlorate. Further elec- trolysis produces more hypochlorite, to which acid may again be added, producing more chlorate. 11 By this means, chlorate can be produced on smooth platinum electrodes with nearly 90 per cent of the theoretical current yield. In place of adding the requisite amount of hydrochloric acid from time to time, the solution may be kept slightly acid by the addition of potassium acid fluoride, KHF1 2 , as patented by the Siemens and Halske Company, 12 or of alkali bicarbonate, which is patented by the Aktiengesellschaft vorm. Schuckert & Co. 13 Oettel found in his early experiments that adding 0.3 gram of potassium hydrate to 100 cubic centimeters of a solution containing 20 grams of potassium chloride does not materially affect the result of the electrotysis, but that as the alkalinity is n Foerster and Miiller, Z. f. Elektroch. 8, 13, (1902). 12 Foerster and Miiller, Z. f. Elektroch. 10, 731, (1904). 1 3 Foerster and Miiller, Z. f. Elektroch. 8, 12, (1902). ELECTROLYSIS OF ALKALI CHLORIDES 93 GRAMS OF N a O H PER LITER FIG. 26. Effect of alkalinity on the elec- trolysis of a solution of sodium chloride increased, the maximum concentration of hypochlorite becomes less, and the principal product of the electrolysis "is chlorate and free oxygen. 14 The curves in Figure 26 15 show the quanti- ties of chlorate and hypochlorife produced per liter by electro- lyzing for one hour solutions containing 200 grams of so- dium chloride and varying quantities of sodium hydrate in one liter. The ordinates are grams of oxygen per liter contained in the chlorate or hypochlorite of the solution, and the abscissae, the number of grains of sodium hydrate added to one liter of the solu- tion. The current density on the anode was 0.04 ampere per square centimeter. This reduction in the hypo- chlorite concentration and increase in that of the chlorate with increasing alkalinity is explained as follows : The reaction by which the chlorate is formed in strongly alkaline solutions is the same as that in neutral or slightly alkaline solutions, and is given in equation (16), and the difference produced by the strong alkalinity is that the chloride finds hydroxyl ions with which to react immediately on the anode, forming hypochlorite. There is therefore a much higher concentration of hypochlorite immediately on the anode than throughout the rest of the solu- tion, and consequently its discharge and the production of chlorate take place when the concentration throughout the solution is very low. 16 When the alkalinity is further in- creased, the hydroxyl ions also begin to be discharged and the yield in chlorate falls below 66.7 per cent, which accounts for the maximum point in the chlorate curve. 14 Z. f. Elektroch. 1, 474, (1895). 15 Miiller, Z. f. Elektroch. 6, 20, (1899) ; Z. f. anorg. Ch. 22, 72, (1900). 16 Foerster and Miiller, Z. f. Elektroch. 9, 182, and 200, (1903) ; also Foerster, Elektrochemie wasseriger Losungeu, p. 366, (1905). 94 APPLIED ELECTROCHEMISTRY Another difference in the electrolysis of strongly alkaline solutions is the effect of temperature. Higher temperature in neutral solutions decreases the maximum concentration of hypo- chlorite obtainable, but in strongly alkaline solutions the effect of temperature is just the reverse, as shown in the curves in Figure 27. 15 The ordi nates are the number of grams of oxygen contained in the hypochlorite or chlo- rate in one liter of a solu- tion originally containing 200 grams of sodium chloride and 40 grams of sodium hydrate the same volume. The 40 TEMPERATURE FIG. 27. Effect of temperature on the electrolysis of an alkaline solution of sodium chloride 111 electrolyses lasted one hour each, with a current density on the anode of 0.045 ampere per square centimeter. Increasing the anode current density tends to counteract this temperature effect. From the explanation given of these curves 17 it does not seem that the effect of temperature in strongly alkaline solutions is thoroughly understood. The Electrolysis of Alkali Chlorides with Platinized Platinum Anodes Lorenz and Wehrlin 1 showed that the use of a platinized platinum anode increases the maximum concentration of hypo- chlorite, and that the oxygen evolution and the production of chlorate do not begin at a time when, on smooth platinum, under the same conditions of the experiment, the oxygen evolu- tion would be considerable. When the electrolysis is continued for a longer time, however, oxygen evolution and chlorate for- mation begin just as on smooth platinum anodes, and according to the same reaction. 2 The only difference is that a higher 17 Foerster and Muller, Z. f. Elektroch. 9, 205, (1903) 1 Z.f. Elektroch. 6, 437, (1900). 2 Foerster and Muller, Z. f. Elektroch. 8, 515, (1902). ELECTROLYSIS OF ALKALI CHLORIDES 95 i 4 80 === GO 40 concentration of hypochlorite is produced before the quantity decomposed in a given time is equal to that produced. This is illustrated by the curves in Figure 28, obtained with a 5.1 normal solution of 10 1 J 4- I L^ c/,T~l 1 1 100 sodium chloride, con- taining 2 grams of potassium chromate per liter. 2 The brok- en curves were ob- tained with a smooth platinum anode, the solid curves with a platinized anode. The ordinates on the right give the per cent of the current yield and the per cent of the current used for the evolu- tion of oxygen, while on the left the ordi- nates give the num- ber of grams per liter of oxygen in the form of chlorate and hypochlorite. The current density on the anode was 0.067 ampere per square centimeter. An explanation of the higher concentration of hypochlorite obtained with platinized anodes will be given below in discussing potentials and decomposition points. FIG. 28. The electrolysis of a 5.1 n. sodium chloride solution, containing 2 grams of potassium chro- mate per liter Dotted lines refer to smooth platinum anode, full lines to platinized platinum anode The Electrolysis of Alkali Chlorides on Carbon Anodes All carbon electrodes are more or less porous ; that portion of their entire volume which consists of pores, or the porosity, varies from 11.2 to 27.8 per cent for different kinds of carbon. For Acheson graphite the porosity is 22.9 per cent. 1 The porosity is calculated from the true and the apparent densities. 1 Foerster, Elektrochemie wasseriger Losungen, p. 372, (1905). 96 APPLIED ELECTROCHEMISTRY The apparent density b is the weight of one cubic centimeter of the material, while the true density a is the weight divided by the volume actually occupied by the material. The value of a is determined by mixing bromoform and chloroform in such proportions that small pieces of the carbon will neither sink nor float when saturated with the mixture. 2 The density of the mixture is then determined by any of the well-known methods, and thus gives that of the carbon directly. The value of the porosity is then 100 - per cent. On dipping a carbon electrode into a solution, the pores become filled with the solution, and the solution contained in the electrode is electrolyzed as well as that on the surface ; but since the dis- solved salt cannot be replaced in the pores as rapidly as in the solution on the surface of the electrode, where stirring replaces the salt decomposed, the solution contained in the pores becomes more dilute than on the surface. Consequently the evolution of oxygen and the production of chlorate will begin sooner, and the maximum concentration of hypochlorite will be less than on a platinum electrode, when the other conditions of the experiment are the same. 3 The effect of changing the chloride concentration or the anode current density on the yield of hypochlorite and on the maximum concentration attainable with carbon anodes is in the same direction as with platinum electrodes. A part of the oxygen liberated oxidizes the carbon to carbon dioxide, part of which, remaining in the solution, makes the solution slightly acid, and therefore changes the hypochlorite to chlorate by equation (16). The formation of carbonic acid takes place in solutions at 20 only to a small extent and after several hours, but at 60 it begins at once, and the total quantities contained in the gases evolved and dissolved in the solution amount to as much as 27 per cent of the amount that would be produced if this were the only product on the anode of the electrolysis. 4 2 Zellner, Z. f. Elektroch. 5, 450, (1899). 8 L. Sproesser, Z. f. Elektroch. 7," 1083, (1901). * Z. f. Elektroch. 7, 944 and 1014, (1901). ELECTROLYSIS OF ALKALI CHLORIDES 97 Carbon anodes are also subject to mechanical destruction, due to crumbling, and in some kinds of carbon this may exceed the loss due to chemical action. The solution in the pores of the carbon may eventually be- come so dilute that hydroxyl ions are discharged, causing the production of hydrochloric acid around the anode ; for hydrogen ions are left behind by the discharge of hydroxyl ions and, com- ing in contact with chlorine ions migrating from the anode, form hydrochloric acid. This fact will be shown later to be of some practical importance. Acheson graphite has been found to last better in the elec- trolysis of chlorides than any other kind of carbon. 1 The Maximum Concentrations of Hypo chlorite and the Maxi- mum Current and Energy Yields of Hypochlorite and Chlorate From what has preceded, it will be evident that the best conditions. for obtaining a high concentration in hypochlorite are to have a neutral, concentrated chloride solution, a low tem- perature, platinized anodes, and to prevent reduction by potas- sium chromate. Column 4 in Table 11 shows the maximum amount of hypochlorite obtainable under different conditions of the experiment. 1 The values given in grams of oxygen may be changed to grams of chlorine by multiplying the former by 5^4^ = 2:22. 2 The solution was 4.79 normal with respect to 16 sodium chloride and contained 2 grams of potassium chromate per liter. In the last experiment the solution was only 1.73 normal. Both electrodes were platinized. Since the decomposition value of a concentrated solution of sodium chloride on either smooth or platinized platinum is 2.2 volts, the minimum amount of energy necessary to produce 1 Foerster and Muller, Z. f. Elektroch. 8, 10, (1902). 2 Foerster and Muller use the ratio 4.44, which is the ratio of the chemical equivalence of the chlorine and oxygen contained in hypochlorite. The ratio of the weights contained, however, is 2.22. 98 APPLIED ELECTROCHEMISTRY one gram of oxygen in the form of hypochlorite is 7.4 watt hours. From the table it is evident that with the lowest current density this value is very closely approached. TABLE 11 GKAMS PER LITER CENTIGRADE PEK SQ. CM. VOLTS Of 0, in OfCljin YIELD IN PER GRAM 2 IN DEOBEES ON ANODE Hypochlo- Hypochlo- PER CENT HYPOCHLORITE rite rite 13 0.017 2.40 4.20 9.3 96 8.4 13 0.017 2.40 5.24 11.6 90 8.95 10 0.07 3.10 6.8 15.0 96 10.84 13 0.17 3.6 5.28 11.7 99 12.2 13 0.17 3.6 8.7 19.8 87 13.5 14 0.17 4.7 5.20 11.5 95 16.6 If chlorate is produced entirely secondarily by acidifying the solution from time to time, no energy is required for its forma- tion beyond the 7.4 watt hours necessary for the production of the hypochlorite. By working in this way and by using plati- nized electrodes, an average current yield of 98 per cent was obtained in a run in which 3.66 volts were applied to the cell. 3 This is 12.5 watt hours per gram of oxygen in the form of chlorate. The current density was 0.117 ampere per square centimeter. By reducing the current density the theoretical value of 7.4 watt hours could of course be more nearly approached. The Production of Per chlorates A perchlorate is a more stable compound than a chlorate, since, as is well known, a chlorate on heating first breaks up into perchlorate, chloride, and oxygen, according to the equations : * 2 KC1O 8 = 2 KC1 + 3 O 2 , (18) 4 KClOg = 3 KC1O 4 + KC1. (19) Foerster and Mtiller, Z. f. Elektroch. 8, 16, (1902). 1 Roscoe and Schorlemmer, Treatise on Chemistry, 1, 235, (1905). ELECTROLYSIS OF ALKALI CHLORIDES 99 A solution of chloric acid is also unstable when its concentra- tion exceeds a certain value, and breaks up as follows : 2 2 HClOg = HC10 4 + HC10 2 . (20) The chloric and chlorous acids then react according to the fol- lowing reversible reaction : HC10 3 + HC10 2 ^ H 2 + 2 C10 2 . (21) These reactions are similar to those by which hypochlorous acid breaks up, 3HC1O = HC1O 3 +2HC1, HC1O + HC1 ^ Cl 2 + H 2 O. Perchlorates are produced in a purely chemical way only by the breaking up of a chlorate, and not by direct oxidation. The electrolytic production of perchlorate and of perchloric acid was discovered by Count Stadion 3 in 1816, but the way in which this oxidation takes place was not understood until recently. This is not a direct oxidation of chlorate to per- chlorate, as would be expressed by the equation : ClOg- + 2 OH- + 2 F = ClO^ + H 2 O, but is due to the discharge of the chlorate ions and their sub- sequent reaction with water, as follows : 4 2 C10 3 - + H 2 + 2 F = HC10 4 + HC1O 2 + O. (22) The oxygen does not escape, but oxidizes the chlorous acid back to chloric acid : HC1O 2 + O = HC10 3 . (23) The principal facts concerning the production of perchlorate are : (1) If the concentration of the chlorate is over 8 per cent, a change in its concentration has no appreciable effect on the current yield ; (2) the yield increases with increasing current density; (3) the yield falls with increasing temperature; (4) platinizing the anode decreases the yield and (5) in 2 Oechsli, Z. f. Elektroch. 9, 807, (1903). 3 Gilbert's Ann. 52, 218, (1816). 4 Oechsli, I.e., p. 819. 100 APPLIED ELECTROCHEMISTRY electrolyzing alkali chlorides, perchlorate is not produced until nearly all of the chloride has been changed to chlorate. In an acid or neutral chlorate solution, perchlorate can be produced with a high current yield, as Table 12 shows, giving the results of an experiment in which 66 per cent sodium chlorate solution was electrolyzed with a smooth platinum anode on which the current density was 0.083 ampere per square centimeter. The temperature was 9 C. TABLE 12 TIME IN MINUTES FROM BEGINNING OF ELECTROLYSIS CURRENT YIELD PER CBNT 5 96.4 20 99.5 35 99.9 180 99.9 210 99.8 300 99.1 Alkalinity prevents the formation of perchlorates ; the cur- rent yield falls to 16 per cent for a solution 0.242 normal with respect to sodium hydrate, with the same current density as in the experiment above. This is probably due to the smaller number of chlorate ions that are liberated as the alkalinity is increased, furnishing hydroxyl ions that are more easily dis- charged than the chlorate. An increase in the current den- sity would be expected to counteract this effect of the alkali, and experiment shows that it does. The lower yield with platinized anodes is due to the lower current density produced by the larger surface. The reduction in the yield by an increase in the temperature is supposed to be due to the greater concentration of hydroxyl ions of water from the increase in the dissociation with the temperature. ELECTROLYSIS OF ALKALI CHLORIDES 101 The Electrolysis of Alkali Chlorides with a Diaphragm If the object in electrolyzing an alkali chloride is to produce an alkali hydrate and chlorine, the anode and cathode must be separated in order to prevent the hydrate and chlorine from mixing. There are four ways in which the separation of the hydrate and chlorine is effected. These are: (1) by the use of a diaphragm ; (2) by inclosing the anode in an inverted, non- conducting bell, with the cathode outside ; (3) by charging a mercury cathode with sodium in an electrolytic cell and decom- posing the sodium amalgam with water in another vessel ; and (4) by a mercury diaphragm, which acts as an intermediate electrode. (1) Since electrolytic conduction takes place through a dia- phragm, it is evident that the separation in this case will not be perfect, for the diaphragm prevents only mechanical mixing. The hydroxyl ions will migrate through the diaphragm and react with the chlorine in the same way as described above. The hydroxyl ions also pass through the diaphragm by ordi- nary diffusion. Electro-osmosis, on the other hand, drives the liquid through the diaphragm from the anode to the cathode, and therefore opposes the diffusion and migration of the hydroxyl ions. 1 If diffusion and osmosis just balance each other, the yield in hydrate can be calculated as follows. 1 Before sodium hydroxide appears at the diaphragm, the sodium chloride transports all of the electricity, but when the hydrate is mixed with the chlorine, the hydrate will also take part in carrying the current through the diaphragm. If the fraction of the current carried by the hydrate is x, that carried by the chloride will be 1 - x, and x and 1 x must be proportional to the conductivities of the hydrate and of the chloride in the solution. If L is the conductivity of the chloride and L 2 that of the hydrate, this is expressed by the equation : 1 X 1 Foerster and Jorre, Z. f. anorg. Ch. 23, 158, (1899). 102 APPLIED ELECTROCHEMISTRY in which c and c 2 are the concentrations in moles per liter, a x and 2 are the dissociations, and A/oc and X"oc are the conduc- tivities at infinite dilution, of the chloride and hydrate respec- X' tively. For potassium chloride and potassium hydrate, ^ = A rji Tf /T n rn\~ i~1 -flJ- i "'VcL MJLl i 7. i i VCL. (^9^ where' <7 clj is the concentration of free chlorine in moles per liter surrounding the anode, tf cl _ is the concentration of chlo- rine ions in the solution, and k is a constant. If <7 C1 and <7 C1 _ RT are both equal, the value of e is - log k, and is called the electrolytic potential. For a solution saturated with chlorine at atmospheric pressure containing 0.064 mole per liter, and normal with respect to chlorine ions, e = 1.667 volts, 1 assum- ing the potential of the dropping electrode to be zero. The negative sign indicates that the solution is negatively charged. Chlorine, cannot therefore be liberated at atmospheric pressure at a potential difference less than this value. On a platinized platinum cathode in an acid solution, normal with respect to hydrogen ions, hydrogen would be liberated at 0.277 volt. But the solution around the cathode is neutral to start with, and soon after the electrolysis has begun is alkaline, due to the i Miiller, Z. f. phys. Ch. 40, 158, (1902). ELECTROLYSIS OF ALKALI CHLORIDES 109 formation of alkali hydrate. The hydrogen ion concentration is then very much reduced below its value in the original neutral solution. This alkalinity might have any value, but for the purpose of calculation the solution around the cathode will be assumed normal, though it would not reach such a high value in a cell not containing a diaphragm. The value of the potential of the cathode on tvhich hydrogen is being liberated would then be 0.54 volt, 0.82 volt more positive than the po- tential in a normal acid solution. 2 The cell would then have an electromotive force of its own of e = 0. 54 - ( - 1. 66) = 2.20 volts. The decomposition point of a concentrated solution of sodium chloride, determined, as usual, with a very small current, is 1.95 volts, but this is because the solution around the cathode is more nearly neutral than assumed above. Continuous elec- trolysis requires from 2.3 to 2.1 volts. 3 1.6 1.8 ANODE POTENTIAL FIG. 31 . Curves showing the relation between current and anode potential, in solutions of sodium chloride and of sodium hypochlorite As was shown above, when the hypochlorite reaches a certain concentration, the hypochlorite ion is also deposited on the anode. It has never been possible to determine the decomposition point of this ion, however. It is evident from the curves in Figure 31, 4 in which the decomposition points of two solutions are 2 Le Blanc, Electrochemistry, p. 209, (1907). Lorenz, Z. f. Elektroch. 4, 247, (1897). * Foerster and Muller, Z. f. Elektroch. 8, 634, (1902). 110 APPLIED ELECTROCHEMISTRY given, one normal with sodium hypochlorite and 0.025 normal with sodium hydrate, the other normal with sodium chloride and 0.01 normal with sodium hydrate, that hypochlorite ions are not liberated before hydroxyl ions. This is shown by the fact that there was no increase in the current below the potential 1.16 volts, approximately the point at which hydroxyl ions are liberated in a normal hydrate solution. It is also evident that the electrolysis of a normal chloride solution begins at a higher potential than the hypocholorite solution. The decom- position point of the hypochlorite ion therefore lies between those of the hydroxyl and the chlorine ions. Since there is a difference of about 0.5 volt between the de- composition points of chlorine and hydroxyl ions, it would seem impossible to liberate chlorine ions in a strongly alkaline solution. This would be the case if it were not that the potential of an anode on which oxygen is liberated increases continuously, and eventually reaches the potential at which chlorine is liberated. If it were not for this in- crease in the potential, caused by the liberation of oxygen, the decomposition of a chloride in an alkaline solution would be impossible. 5 Another effect which tends to make chlorine de- posit in an alkaline solution is the fact that the hydrate has a depolarizing effect on the chlorine, in consequence of which chlorine will be liberated at a lower potential than that neces- sary for its deposition at atmospheric pressure. Table 14 shows TABLE 14 PER CENT YIELD IN ACTIVE OXYGEN POTENTIAL AMPEBES Total As Hypo- chlorite As Chlorate -1.09 - 1 0.0 -1.21 to -1.27 0.016 0.9 - 1.30 to - 1.51 0.28 to 0.14 3.2 2.6 0.6 - 1.51 to - 1.595 0.5 to 0.4 16.4 8.2 8.2 6 Foerster and Mtiller, Z. f. Elektroch. 9, 184, (1903). ELECTROLYSIS OF ALKALI CHLORIDES 111 that hypochlorite and chlorate are formed in a solution normal with sodium hydrate, and 3.6 normal with sodium chloride, at an anode potential below 1.667 volts, the potential at which chlorine is liberated at atmospheric pressure. 6 The anode was platinized platinum, of 14 square centimeters area. It will be noticed that as the anode potential increases in nu- merical value, the proportion of chlorate to hypochlorite in- creases. This is due to the fact that the hypochlorite ions, which are more easily discharged than the chlorine ions, are more subject to deposition as the potential of the anode in- 1.1 1.3 ' 1.5 1.7 1.9 2.1 ABSOLUTE POTENTIAL DIFFERENCE BETWEEN ANODE AND SOLUTION FIG. 32. Curves showing the relation between current and anode potential for smooth and for platinized platinum anodes creases, with the subsequent production of chlorate according to equation (16). On smooth platinum anodes the potential difference during electrolysis is about 0.58 volt greater than on platinized plati- num. 7 The decomposition points of sodium chloride on plati- e Foerster and Miiller, Z. f . Elektroch. 9, 183, and 201, (1903). 7 Z. f. Elektroch. 6, 437, (1900). 112 APPLIED ELECTROCHEMISTRY nized platinum and on smooth platinum anodes shows the same difference, as is seen from the curves in Figure 32. 8 It is evi- dent that the overpressure of an anion is a function not only of its own chemical nature, but also of the solution from which it is deposited, of the current density, and of the material com- posing the anode. The cause of this overpressure of 0.58 volt on platinum is not well understood; it may be due to the resistance of a film of gas liberated on the anode. There is a corresponding overvoltage in other solutions, such as sodium hydrate and sulphuric acid, where oxygen, in place of chlorine, is liberated. These over- pressures are not equal for the same current density in these different solutions. 9 Though the overpressure on smooth platinum anodes may not itself be understood, its presence offers a possible explanation of the higher concentration of hypochlorite obtained with a platinized anode, for the relation between the decomposition potential and the concentration of ions is that the decomposition potential decreases as the concentration increases. Therefore, with a lower anode potential, the concentration of the hypo- chlorite ions would have to be greater before decomposition takes place. 10 It is an experimental fact, as has been stated above, that very little perchlorate is produced until most of the chloride has been changed to chlorate. This is due to the fact that the decom- position potential of normal sodium chlorate is 2.36 volts, 11 while that of the chloride is 1.95 volts. 3 The high potential re- quired for the deposition of the chlorate cannot therefore be reached until most of the chloride has been used up. When chlorine- is dissolved in water, according to equations (2) and (4), a certain amount of hypochlorous acid and hypo- chlorite will be produced. Both hypochlorous acid arid hypo- chlorite are oxidizing agents, and therefore give an unattackable s Miiller, Z. f. Elektroch. 8, 426, (1902). Foerster and Miiller, Z. f. Elektroch. 8, 533, (1902). i Foerster and Miiller, Z. f. Elektroch. 9, 199, (1903). 11 Wohlwill, Z. f. Elektrocb. 5, 52, (1898). ELECTROLYSIS OF ALKALI CHLORIDES 113 electrode a definite potential. If the reactions by which they give off oxygen, or what is the same thing, hydroxyl ions, are HOC1 = OH- + Cl- + 2 F, (30) CIO- + H 2 O = Cl- + 2 OH + 2 F, (31) the potentials would be given by the equations RT , (33 > * V OH- and for equilibrium concentrations, tfj being taken from equation (29). When chlorine is liberated on an unattackable anode, the equilibrium represented by (9), = 3.6 x IP" 11 " ocl = 1.4 xlO' 17 must be established, and, assuming the chlorine electrode is re- versible, the production of hypochlorite and hypochlorous acid must be, according to (31) and (32), taken from right to left. This means that a primary production of hypochlorite and hypochlorous acid takes place on the anode to a small extent. Fluorides, Bromides, and Iodides The electrolysis of the other alkali halogen compounds has not attained anything like the commercial importance of the electrolysis of chlorides ; still, for the sake of completeness, the behavior of the other alkali halides on electrolysis will be briefly described. Fluorine decomposes water with the evolution of oxygen and ozone : 2 OH- + 2 Fl- = H 2 O + O. (35) No oxygen compounds of fluorine are known, consequently the electrolysis of fluorides offers nothing to compare with what is obtained in the case of chlorides. 114 APPLIED ELECTROCHEMISTRY Bromine enters into exactly similar equilibria when added to alkali hydrate to those already described in the case of chlorine. They are represented by the equations : l Br 2 + OH- = HOBr + Bi-1 HOBr + OH- = BrO- + H 2 OJ ' Hypobromite is therefore always the first product of the reac- tion when bromine acts on alkali hydrate. When one mole of bromine acts on one equivalent of hydrate, the reaction is not as complete as in the case of chlorine, but appreciable quantities of bromine and hydrate remain unchanged. The formation of bromate according to the equation 2 HOBr + NaBrO = NaBrO 3 + 2 HBr (37) takes place with over 100 times the velocity of the correspond- ing reaction for chlorate. This reaction takes place even in slightly alkaline solutions with a high velocity, on account of the greater hydrolysis of hypobromite, but in solutions that are at least 0.1 normal with respect to hydrate, the hydrolysis has been so far reduced that hypobromite is as stable as hypo- chlorite. When a concentration of the hydrate is still further increased, the rate at which bromate is produced increases, probably according to the reaction : 3 NaBrO = NaBrO 3 + 2 NaBr. (38) This differs from the corresponding reaction for chlorate, in that it proceeds with scarcely any evolution of oxygen. This reaction, however, is very much slower than that represented by equation (37), and need not be considered in the practical preparation of bromate. In electrolyzing a bromide solution, free bromine is liberated on the anode, accompanied by oxygen from the discharge of hydroxyl ions, and produces hypobromite with the hydrate formed at the cathode. The concentration of the hypobromite increases up to a certain point, after which it remains constant, and the only product of the electrolysis is then bromate. As i Horst Kretzschmar, Z. f. Elektroch. 10, 789, (1904). ELECTROLYSIS OF ALKALI CHLORIDES 115 the hypobromite increases in concentration, the evolution of oxygen also increases, the hydroxyl ions for which are fur- nished by the hydrolysis of the hypobromite. Bromate is formed partly by the secondary oxidation of hypo- bromite by hypobromous acid, which is always present to a certain extent on the anode, and partly by direct oxidation according to the equation : (39) The hypobromite ion is not discharged, so there is no reaction between it and water, as there is in the case of the hypochlorite ion. The concentration of hypobromite attainable is greatest with a high current density, a high concentration of bromide, and a low temperature. It is also higher on platinized anodes than on smooth, as is the case with hypochlorite. The highest con- centration of hypobromite attainable is about the same as that of hypochlorite, but the current yield is less, on account of the greater tendency to form bromate. Unless potassium chro- mate is added to the solution, bromate, as well as hypobromite, is subject to reduction on a smooth platinum cathode, 2 which is another point of difference between chlorate and bromate. Perbromic acid and its salts cannot be produced by elec- trolysis, and it is doubtful whether they exist at all. 3 When iodine is brought in contact with hydrate, the equilibria I 2 +OH-=HOI + I- 1 HOI + OH- = 01- + H 2 I are established exactly as in the case of chlorine and bromine. 4 Hypoiodite is very considerably hydrolyzed, and therefore, unless the solution is very alkaline, it changes rapidly to iodate by the reaction : 2 HOI + KIO = KIO 3 + 2 HI. (41) 2 H. Pauli, Z. f. Elektroch. 3, 474, (1897). 8 Roscoe and Schorlemmer, Treatise on Chemistry, 1, 358, (1905). 4 Foerster and K. Gyr, Z. f. Elektroch. 9, 1, (1903). 116 APPLIED ELECTROCHEMISTRY If an excess of alkali is present, however, the hydrolysis is driven back, and hypoiodite can be obtained free from iodate. The formation of iodate is accelerated by an increase in the temperature and concentration of the iodide, and by decreasing the alkalinity. The rapidity with which hypoiodite changes to iodate is shown by the following facts : If 50 cubic centimeters of a 0.1 normal iodine solution are mixed with 50 cubic centimeters of a normal sodium hydrate solution at zero degrees, a 0.05 normal hypoiodite solution would be 100 per cent yield. Imme- diately after mixing, however, there is only 95 per cent of this amount of hypoiodite, and after 2 minutes, only 75 per cent remains. On dilution it is more stable ; a 0.01 normal hypoio- dite solution remains practically unchanged for a few minutes in a 0.1 normal alkaline solution at room temperature. On electrolyzing a neutral solution of alkali iodide, 5 the io- dine liberated on the anode comes in contact with the hydrate from the cathode, and the first product is hypoiodite. This changes over to iodate rapidly, as shown above, even in an alka- line solution, so that the electrolysis of an alkali iodide solution is similar to that of a slightly acid chloride solution. Conse- quently the hypoiodite solution reaches a limiting concentra- tion, after which the product of the electrolysis is exclusively iodate. This limiting concentration of hypoiodite is determined by the current density, temperature, and the concentration of iodide and alkali. An increase in the alkalinity increases the limiting concentration of the hypoiodite, while it decreases that of the hypochlorite. This is due to the different ways in which iodate and chlorate are formed in alkaline solutions. As the hypoiodite never can become concentrated, the possi- bility of the electrolytic discharge of the hypoiodite ion is very small. Therefore the oxygen evolution, which takes place only when the iodide is dilute and the solution is alkaline, must be due nearly entirely to the discharge of hydroxyl ions. It is therefore in no way connected with the formation of io- date. 6 Foerster and Gyr, Z. f. Elektroch. 9, 215, (1903). ELECTROLYSIS OF ALKALI CHLORIDES 117 Periodates cannot be produced by the electrolysis of iodates except with a diaphragm. 6 This is shown by the fact that without a diaphragm no hydrogen is evolved on electrolysis, but is all used in reducing the iodate. After electrolysis has proceeded a while, the oxygen evolution also becomes zero. There is then a constant amount of iodide and iodate in the solution ; as fast as iodate is formed on the anode, it is reduced on the cathode. In neutral solutions iodate is not oxidized to periodate, and in alkaline solutions, potassium chromate does not prevent the reduction of iodate to iodide. By using a diaphragm, a current yield in periodate of about 26 per cent can be obtained. The best conditions are low tem- perature, low-current density, and at least 4 per cent alkalinity. 2. TECHNICAL CELLS FOR HYPOCHLORITE, CHLORATE, HY- DRATE, AND CHLORINE Hypochlorite. Hermite's cell, patented in 1887, was the first cell to be even moderately suc- cessful for the electrolytic manufac- ture of hypochlorite. 1 It consisted of a rectangular box of ceramic with a grooved channel around the top for carrying off the solution of sodium and magnesium chlorides, which entered at the bottom. The cathode consisted of numerous disks of zinc supported on two slowly rotating shafts running through the box and separated from each other by a parti- tion. The anodes, consisting of thin sheets of platinum held on a noncon- ducting frame, were placed between the zinc disks. In practice this cell FIG. 33. Elevation of Kellner cell 6 E. Muller, Z. f. Elektroch, 7, 509, (1901). 1 W. H. Walker, Electroch. Ind. 1, 440, (1903); Engelhardt, HypochJorite and Elektrische Bleiche, p. 77, (1903). 118 APPLIED ELECTROCHEMISTRY g.ives a current yield of about 40 per cent and an energy yield of one kilogram of chlorine for twelve kilowatt hours. 2 The Kellner cell, made by the Siemens and Halske Company, is shown in Figures 33, 34, and 35. A glazed stoneware vessel is divided into a number of compartments by glass plates fitted into grooves in the sides of the cell. The plates are wound with platinum -iridium wire, which acts as intermediate electrodes, form- [f] [i fj H H'H \] 1} fi H] ing the anodes on one side and the cathode on the other side of of the glass plates. The solution enters through holes in the bottom of the cell and the electrolyzed solution flows out spouts at the FIG. 34. Electrodes Kellner cell a FIG. 35. Plan of Kellner coll FIG. 36. Kellner cell top into a vessel containing a cooling coil. From here it is pumped up through the cell again. This circulation continues 2 Engelhardt, I.e. p. 86. ELECTROLYSIS OF ALKALI CHLORIDES 119 until the desired strength of hypochlorite has been obtained. This is illustrated in Figure 36. The Schuckert cell is also made by the Siemens and Halske Company. It is made of stoneware and is divided into eight FIG. 37. Horizontal section of Haas aud Oettel cell or ten compartments, each having two graphite cathodes and a Pt-Ir foil anode. The solution enters at one end and travels in a zigzag direction through the different compartments. Each cell has a cooling coil, and no pumps are needed for circulation. The units are built in pairs and are designed for 110 volts. 120 APPLIED ELECTROCHEMISTRY The Haas and Oettel cell is shown in horizontal and vertical cross sections in Figures 37 and 38. 3 The electrolyzer b is im- mersed in the solution in the storage vessel a. The electrolyzer consists of a vessel divided into several compartments c by divisions r, made of carbon or any suitable material, and form- ing the intermediate electrodes. The liquid enters the elec- trolyzer through the passage d, one of which leads into each compartment. As soon as the current is turned on, gas is pro- duced in each compartment, which rises and, carrying the liquid with it, causes it to flow through the channels e, as shown by the arrows. This automatic circulation is very efficient. A cooling coil in the container prevents the temperature from rising too high. The electrolysis is continued till the concen- tration of the hypochlorite has reached the desired value. Ibsss^^^^^ 1" FIG. 38. Vertical section of Haas and Oettel cell This cell was never put on the market in this country in the form shown, 4 but an improved cell is made by the National Laundry Machinery Company of Dayton, Ohio, the details of which are not now available. Among a number of other factors, the cost of the production 8 U. S. Pat. 718,249, (1903). 4 Communication from the National Laundry Machinery Company. ELECTROLYSIS OF ALKALI CHLORIDES 121 of hypochlorite depends on the cost of salt and of power, and on the concentration of the hypochlorite produced ; for, as was shown above, the current efficiency of the production of hypo- chlorite approaches zero as the concentration increases. For cotton bleaching the hypochlorite is diluted to three grams of active chlorine per liter, and is discarded after using. 5 Less salt will therefore be lost if as much as possible is changed to hypochlorite, but the cost of power increases as the concentra- tion increases. The concentration to which it will be most economical to continue the electrolysis will therefore depend on the relative cost of power and of salt, assuming all other conditions of the experiment constant. There will then be a concentration of hypochlorite for which the cost will be a mini^ mum, assuming a definite cost for the salt and the power. This minimum cost is found by plotting as ordinates the cost TABLE 15 The Kellner Cell ORIGINAL NaCl CONC. KG. PER 100 L. AMPERES PER CELL GRM. ACTIVE C1 2 PER L. PER CENT CURRENT YIELD KW. HR. PER KG. ACTIVE Cl, KG. SALT PER KG. ACTIVE 01, 6.3 120 1.84 77.8 5.9 34.2 6.3 120 3.34 68.8 6.7 18.9 6.3 120 7.06 61.8 7.4 8.9 6.3 120 10.01 44.5 10.3 6.3 10.0 137 3.09 81.5 5.2 32.4 10.0 137 6.85 68.6 6.2 14.6 10.0 137 10.44 58.2 7.3 9.6 10.0 137 12.96 45.9 9.2 7.7 15.0 126 3.00 82.3 5.1 50.0 15.0 126 6.28 73.0 5.7 23.9 15.0 126 10.50 65.3 6.4 14.3 15.0 126 13.50 54.2 7.7 11.1 20.0 130 2.48 90.1 4.5 80.7 20.0 130 6.58 78.0 5.2 30.4 20.0 130 10.09 70.0 5.8 19.8 6 W. H. Walker, Trans. Am. Electrochem. Soc. 9, 23, (1906). 122 APPLIED ELECTROCHEMISTRY of power for a definite amount of hypochlorite at different con- centrations, and also as ordinates the cost of the salt required for the different concentrations of hypochlorite. The curve repre- senting the sum of these costs will be found to have a minimum value. Table 15 gives some data on the yield of active chlorine in the Kellner cell, taken from cells in actual operation. 6 The yields of active chlorine in the Haas and Oettel appara- tus are given in Table 16 7 TABLE 16 The Haas and Oettel Cell GEM. ACTIVE Clj PER L. PER CENT CURRENT YIELD KW. HR. PER KG. ACTIVE C1 2 KG. SALT PER KG. ACTIVE C1 2 2.55 95.0 3.31 66.6 4.59 82.4 3.82 37.0 8.82 64.8 4.85 19.3 12.30 56.7 5.54 13.8 14.31 52.8 5.96 11.9 The yields in active chlorine for the Schuckert cell are given in Table 17. 1 TABLE 17 The Schuckert Cell ORIGINAL NaCl CONC. PER CENT GRM. ACTIVE C1 2 PER L. KW. HR. PER KG. ACTIVE C1 2 KG. SALT PER KG. ACTIVE C1 2 10 10-22 7 5- 5.3 10 16 6 6- 6.5 10 10-12 5 10-10.6 15 20-22 5.5-6 7.5- 8 15 10-12 4.5-5 15-16 6 Englehardt, I.e. p. 158. 7 Oettel, Z. f. Elektroch. 7, 315, (1900). ELECTROLYSIS OF ALKALI CHLORIDES 123 Chlorate Cells. Since chlorate is made directly from hypo- chlorite, a chlorate cell would not be expected to differ from a hypochlorite cell in any mechanical details. The earlier chlo- rate cells, however, contained a diaphragm, and the cathode solution was allowed to circulate to the anode compartment. This was to prevent the reduction of the hypochlorite from which the chlorate is produced ; but since the discovery of the action of potassium chromate, reduction can be avoided with- out a diaphragm. The first process to be used in practice was that of Gall and Montlaur, patented in 1884. 8 This cell originally contained a diaphragm to prevent reduction, and the solution circulated from the cathode to the anode by means of external pipes. The solution must, of course, leave the anode compartment as rapidly as it flows in, but whether it leaves the cell entirely or returns to the cathode compartment is not stated. Since 1897 the diaphragms have been given up. A plant employing this pro- cess was put in operation at Vallorbe in 1891, and another in St. Michel, Savoy, in 1896. Very little information concerning these plants has been published. In 1892 the National Electrolytic Company at Niagara Falls employed the chlorate cell of W. T. Gibbs. 9 A number of these cells clamped together are shown in Figure 39, and a side elevation of one cell on the line 22 of the preceding figure, in Figure 40. 10 Each cell consists of a frame A made of wood with a metallic resistant lining B. The rods forming the cathode are attached on one side of this frame, and on the other, the anode, consisting of a metallic plate D faced with platinum E. Copper is preferred for the cathode and lead for the plate D. Successive frames are separated from each other by gaskets F. G- are supply tubes and .ff are vents for the escape of gas and liquid. The cells are clamped together by the plates JK and the bolts L. Each pair of electrodes is separated by the corre- sponding gasket. The horizontal insulating rods prevent 8 J. B. C. Kershaw, Die Elektrolytische Chloratindustrie, p. 19, (1905). 9 J. W. Richards, Electrochern. Ind. 1, 19, (1902). 10 TJ. S. Pat. 665,426, (1901). 124 APPLIED ELECTROCHEMISTRY short circuit between the anodes and cathodes, which are only from 1 to 3 millimeters apart. The electrolyte circulates from the cell to a cooling vessel where the chlorate is precipitated. FIG. 39. Gibb's cells clamped together More chloride is then added, and the solution is returned to the electrolyzing cell. A convenient size for these cells is 65 by 4-3 centimeters and 7.5 centimeters thick. ELECTROLYSIS OF ALKALI CHLORIDES 125 The cell of Lederlin and Corbin, used at Chedde, is of the open type. 11 It contains a platinum anode and two cathodes of copper, bronze, brass, or iron. The anode has an area of 10 square centimeters and the cathode, 32. FIG. 40. Section of single Gibb's cell The chlorate is generally purified by recrystallization, and the recrystallizing apparatus is an important part of a chlorate plant. The yield at Vallorbe was at first 55.9 grams per kilowatt hour, though this has since been considerably increased. 12 11 Kershaw, I.e. p. 38. Kershaw, I.e. p. 126 APPLIED ELECTROCHEMISTRY The yield obtained at Chedde with the Lederlin and Corbin cell in a slightly acid solution containing potassium bichromate was 0.69 gram per ampere hour, or 90 per cent of the theoretical. Perchlorates. The cells used for the production of chlorates can be used equally well for perchlorates. Whether there is a difference in practice cannot be stated, for no description of a perchlorate cell has been published. Alkali Hydrates and Chlorine. In cells in which hydrate and chlorine are to be the final product, the anode must be sep- arated from the cathode so that the chlorine and hydrate can- not mix. In the first type of cell to be considered, this is accomplished by means of a porous diaphragm. A very large number of such cells have been patented, but only a few need be described. One of the simplest of the diaphragm cells is McDonald's, used by the Clarion Paper Mill at Johnsonburg, Pennsylvania, 13 and the United States Reduction and Refining Company in Colorado. At the latter plant, there are 75 cells, producing 1500 pounds of chlorine in 24 hours. 14 Two vertical sections of LENGTHS SECTION. CROSS SECTION. BBBBSBBBB FIG. 41. McDonald cell the cell are shown in Figure 41. It consists of a cast-iron tank, 1 foot wide, 1 foot high, and 5 feet and 2 inches long, with two longitudinal perforated partitions. The perforations are % inch in diameter, and there are 4 or 5 to the square inch. A diaphragm is placed next each partition in the middle com- 13 Electrochem. Ind. 1, 387, (1903). 14 J. B. procker, Electrochem. and Met. Ind. 5, 201, (1907). ELECTROLYSIS OF ALKALI CHLORIDES 127 partment, containing the anode. The diaphragms consist of asbestos paper fastened to asbestos cloth by sodium silicate, and are held in position by cement placed over both end walls and the bottom of the anode compartment. This compartment is closed by a cast-iron cover 5 inches deep, 6 inches wide, and nearly 5 feet long, into which the anodes are cemented. It is lined with cement and painted inside with asbestos varnish. The anode cohsists of blocks of graphitized carbon, 4 inches square and 10 inches long, into each of which a copper rod is fastened by lead for the electrical connection. The partition walls form the cathode. The partition walls are flanged, forming a seat to hold the cover, which is surrounded by a layer of cement. The chlorine is conducted from the anode compartment by a lead pipe to a gas main which leads to absorbing towers containing lime- water. Brine circulates through the anode compartment. The diaphragms last about 8 months, 16 after which time the pores become clogged. The sodium hydrate solution leaving the cathode compart- ment contains from 7 to 18 per cent sodium hydrate. When the diaphragm is new, the level of the liquid in the anode arid cathode compartments is nearly the same, but when it becomes more or less stopped up, the depth of the liquid -in the cathode compartment is only an inch or two. The Hargreaves-Bird cell consists of a cast-iron box 10 feet in length, 14 inches in width, and 5 feet in height. 16 It is divided into three compartments by two diaphragms made on heavy copper gauze, which forms the cathode. The space between the diaphragms is the anode compartment, through which brine circulates. There is no liquid in the anode com- partment except what percolates through the diaphragm. Steam and carbonic gas are blown through the two outer com- partments and change the hyrate formed on the outside of the diaphragm to sodium carbonate. This cell takes 2000 amperes at from 4 to 4.5 volts. The anode is a row of gas carbons, is L. Rostosky, Z. f. Elektroch. 11, 21, (1905). is Electrochero. and Met. Ind. 3, 350, (1905). 128 APPLIED ELECTROCHEMISTRY which last 30 to 40 days. The diaphragms last about the same length of time. The Hargreaves-Bird cell is shown in Figure 42, which is a partial longitudinal section and side elevation, and in Figure 43, which is a section perpendicular to the length. 17 The outside frame I is of iron lined with cement and bricks w, which are saturated with tar to prevent leakage.- The space FIG. 42. Hargreaves-Bird cell, side elevation / is the anode compartment through which the chloride solution circulates, entering through the pipe g and leaving through h. The diaphragms are made of asbestos paper, the pores of which have been filled with hydrated silicate of lime or magnesia. 18 In the cathode chamber a number of copper strips b are placed, imbedded in cement e, extending from the cover plate c to the cathode c?, and inclined downwards. These direct the 17 U. S. Pat. 655,343, (1900). U. S. Pat. 596,157, (1897). ELECTROLYSIS OF ALKALI CHLORIDES 129 densed vapor against the cathode to wash away the alkali as it is formed. The lower edges of the strips have openings, in order to allow the steam and gas to pass freely over the cathode, a', a r are the injectors for supplying carbonic acid gas and steam to the cathode chambers. Z 2 , Z 2 are pipes for draining the cathode chambers. The chlorine passes from the anode chambers to the towers, where it is absorbed by milk of lime. The West Virginia Pulp and Paper Company, Me- chaiiicsville, New York, use this cell for making their bleaching solu- tions. 19 This plant consists of two rows of 14 cells each, n LJ\ FIG. 43. Harsreaves-Bird cell, end section all connected in series. Perhaps the most efficient diaphragm cell in use is the Townsend cell, repre- sented in cross section in Figure 44, and in perspective in Figure 45. 20 The anode space is inclosed between a lid (7, two vertical diaphragms D, and a non- conducting body H. Graphite anodes pass through the lid on the cell. The perforated iron cathode plates S are in close contact with the diaphragms. These plates are fastened to two iron sides J, which form the cathode compartment. The anode compartment is filled with brine T, and the cathode corn- is Electrochem. and Met. Ind. 6, 227, (1908). 20 Electrochem. and Met. Ind. 5, 209, (1907). FIG. 44. Townsend cell 130 APPLIED ELECTROCHEMISTRY partment with kerosene oil K. The brine percolates through the diaphragm, and, when the current is turned on, it contains hydrate. The aqueous solution, on passing the dia- phragm, comes in contact with the kerosene and forms drops which fall to the bottom of the compart- ment, are collected in the pocket A, and are drained off through P. The solution leaving P contains 150 grams of sodium hydrate and 213 grams of salt per liter. The salt is separated by evaporation and is used over again. The continual percolation prevents nearly all diffusion of hy- drate back to the anode. The rate of percolation for a 2500- ampere cell is from 15 to 30 liters an hour. 21 The Town send cell is 8 feet in length, 3 feet in depth, and 1 foot in width, and consists of a U-shaped concrete body against which the two iron side plates are clamped. A rubber gasket is placed between the concrete and the iron to make a tight joint. Brine circulates through the anode compartment, and during its passage the specific gravity falls from 1.2 to 1.18. On leaving the cell it is resaturated and is then ready to be passed through again. There is a loss in kerosene which amounts in cost to about two dollars a day for a large plant. The diaphragms of the Townsend cell consist of a woven 21 Baekeland, Electrochem. and Met. Ind. 7, 314, (1909). FIG. 45. Townsend cell ELECTROLYSIS OF ALKALI CHLORIDES 131 sheet of asbestos cloth, the pores of which are filled with a mix- ture of iron oxide, asbestos fiber, and colloid iron hydroxide. This mixture is applied with a brush like ordinary paint. Di- aphragms may be renovated by scrubbing and washing the surface with water, allowing to dry, and repainting with this mixture. This operation is not. necessary more than once in five weeks, and sometimes not for several months. The current efficiency of the Townsend cell is as high as 96 or 97 per cent under ordinary conditions, with a current density on the anode of 1 ampere per square inch and about 4 volts on each cell. 22 This cell has been in use at Niagara Falls in the plant of the Development and Funding Company since 1906. This plant originally consumed 1000 kilowatts, and according to latest accounts it was being increased to four times this capacity. 21 Not much information concerning the bell process as actually arranged in practice is available. The process is carried out by the Oesterreiche Verein fur Chemische und Metallurgische Production in Aussig, and at several places in Germany. Fig- ure 46 shows FIG. 46. Cell for Bell process two cross sections of the cell, 25 of which are placed side by side in each bath. 23 The solution leaving the bath is said to contain 100 to 150 grams of alkali hydrate per liter, at a cur- rent yield of 85 to 90 per cent and with 4 to 4.5 volts per cell. 24 The Castner cell 25 is represented in Figure 30. It is a slate box 4 feet square, and 6 inches deep, the joints of which are 22 For laboratory tests on the efficiency of this cell, see Richardson and Patter- son, Trans. Am. Electrochem. Soc. 7,311, (1910). 28 Z. f. Elektroch. 7, 925, (1901). 24 Haeussermann, Dinglers polyt. J. 315, 475, (1900). 25 U.S. Pat. 528,322, (1894). 132 APPLIED ELECTROCHEMISTRY made tight with rubber cement. 26 Two partitions, reaching to within ^g inch of the bottom, divide the cell into three compart- ments. The two outside compartments contain the graphite anodes A, and the middle compartment, the iron cathode 0. Brine circulates through the anode compartments, and pure water is supplied to the cathode compartment. The cell is piv- oted on two points at one end and the other is raised and lowered FIG. 47. Whiting electrolytic cell, plan about J inch once a minute, causing the mercury to circulate be- tween the anode and cathode compartments. The hydrate leaving the cathode compartment has a specific gravity of 1.27. This is evaporated to solid hydrate in large iron pans. Each cell takes about 100 pounds of mercury, which is a very large item of expense. The current for each cell is 630 amperes at 4.3 volts, and the current efficiency is about 90 per cent. ae J. W. Richards, Electrochem. Ind. 1, 12, (1902). ELECTROLYSIS OF ALKALI CHLORIDES 133 The Whiting mercury cell is 27 of a different type from the Castner cell. The sodium is not electrolyzed out of the amal- gam, but the amalgam is withdrawn from the electrolyzing chamber and treated with water in a decomposing chamber where the hydrate is formed. A number of electrolytic com- partments are placed in parallel and are operated successively. B "b FIG. 48. Whiting electrolytic cell, cross section so that the cell is continuous in its action, though intermittent in principle. This cell, shown in Figures 47, 48, and 49, is a massive con- crete structure supported on four concrete pedestals, from which it is insulated. It consists of a shallow box divided into two compartments, A and B, by a concrete partition. The bottom of the decomposing chamber is divided by low glass partitions into a number of sections having V-shaped bottoms ^ Jasper Whiting, Trans. Am. Electrochem. Soc. 17, 327, (1910). 134 APPLIED ELECTROCHEMISTRY sloping at a slight angle towards the central slot D. These slots lead through the concrete partition into the oxidizing chamber B, where they turn upward and are closed by valves E. The valves are operated by the cams F, which are attached to a slowly revolving shaft Gr. The other ends of the slots are connected by the channel H, called the distributing level. This connects with a secondary channel /, which leads through one of the side walls of the cell to a pump J", at the extreme end of the oxidizing compartment. Mercury covers the bottom of the decomposing compartment, filling the above-described sections to a common level. The anodes K are slabs of Acheson graphite, perforated to allow the chlorine to escape, and rest upon the ledges L, placed at the ends of the section in FIG. 40. Whiting electrolytic cell, longitudinal section such a way as to make a very short distance between the anode and the mercury cathode. The anodes are connected to the dynamo by the leads M. The oxidizing chamber is divided into three compartments P, lined with graphite and sloping downward in successively opposite directions, forming a zigzag path to the pump pit Q, where the stoneware rotary pump J is placed. Brine fills the ELECTROLYSIS OF ALKALI CHLORIDES 135 decomposing chamber, and water or alkali hydrate fills the oxi- dizing chamber. The action is as follows : The floor of several sections of the decomposing chamber is covered with mercury, maintained at a common level by the distributing level. The current flows from the anode through the brine to the mercury and out by the iron rods R, partly imbedded in the concrete. When the electrolysis has proceeded about two minutes, the valve at the point of exit of one of the sections is opened by the action of the cam, and the entire mass of sodium amalgam in the section sinks into the slot and through the connecting pipe into the oxidizing chamber. When the mercury is out of the cell, the valve is closed by the cam. Mercury free from sodium then flows into the empty chamber by way of the distributing level, until the common level is reached. In the meantime the sodium amalgam in the oxidizing chamber flows by gravity over the graphite plates P to the pump pit. On reaching this point the mercury has been deprived of its sodium, and is raised by the pump into the wall pipe of the decomposing chamber, completing the cycle. The brine is fed in between the electrodes from the recep- tacles S, equal in number to the sections of the decomposing chamber. They are formed in the cover of the decomposing compartment, and are connected by a channel T. Glass tubes lead from the bottom of the receptacle S through the anode and terminate below the surface of the mercury near the middle of each section. As long as the sections are filled with mer- cury the lower ends of the tubes are sealed, but when the mercury is drawn off, a definite quantity of concentrated brine flows into the section. The graphite slabs in the oxidizing chamber contain a large number of channels through which the mercury flows. The sides of the channels extend into the caustic solution and form the cathode of a short-circuited couple. It is difficult to main- tain good contact between the graphite and mercury on account of the hydrogen evolved, but this difficulty was overcome by boring holes j- inch deep and J inch in diameter at frequent 136 APPLIED ELECTROCHEMISTRY intervals in the channels, and filling them with pure mercury at the start. This mercury remains pure and makes good con- tact with the amalgam and the graphite. The cell used at the Oxford Paper Company's works in Rum- ford, Maine, is 1.8 meter square. It consists of five sections and takes a current of from 1200 to 1400 amperes at 4 volts. This corresponds to an anode current density of 11 amperes per square decimeter. The current efficiency is from 90 to 95 per cent. The temperature is about 40 degrees. Each cell requires from 350 to 375 pounds of mercury. A 20 per cent hydrate solution is obtained, though one with 49 per cent can be made if desired. The chlorine gas is 98 per cent pure, the remaining 2 per cent being hydrogen. CHAPTER VII THE ELECTROLYSIS OF WATER HYDROGEN and oxygen have a number of technical applica- tions that require their manufacture on a large scale. Such uses are welding with the oxyhydrogen flame, as is done in joining the lead plates of storage batteries; hydrogen is used for filling balloons, and oxygen is used for chem- ical and medicinal purposes. Hydrogen and oxygen are produced on a commercial scale by the electrolysis of aqueous solutions, and of course the object of the large number of patents taken out in this field is to keep the hydrogen and oxygen separate from each other. For this purpose the anode and cathode compartments have to be sep- arated by a partition of some kind. The different methods of separating the gases will be Schnitt FIGS. 50-53. Schmidt's apparatus for the electrolysis of water illustrated in the description of the following cells. The cell designed by Dr. O. Schmidt 1 is shown in sections in Figures 50-53, and a general view in Figure 54. It consists 1 Engelhardt, Die Elektrolyse des Wassers, p. 24, (1902); Z. f. Elektroch. 7, 294, (1901). 137 138 APPLIED ELECTROCHEMISTRY of a number of iron plates e having thick rims and separated by diaphragms d. These plates are the cathode in one cell and the anode of the following cell. Each plate has two holes in the thick rims h,o and w, w' , so that the apparatus is traversed above and below by two canals. The lower canals are for sup- plying the water as it is decomposed, and the upper are for al- FIG. 54. Schmidt's apparatus for the electrolysis of water lowing the gases to escape. The canals w and h connect with the cathode chambers, w' and o with the anode chambers. The two canals for adding water, w and w', are connected with a common filling tube W by the pipes w 2 , up, and at the other end of the apparatus the two gas canals connect with reservoirs .ZTand 0, where the gas is separated from the liquid carried along with it. The liquid then returns to its respective chamber in the electrolyzer. The stopcock a is for emptying the apparatus. The diaphragms are of asbestos with rubber edges to prevent leakage. The electrolyte is a dilute solution of potassium car- bonate. Each cell has 2.5 volts impressed, and the current yield is nearly 100 per cent. The oxygen is on the average THE ELECTROLYSIS OF WATER 139 97 per cent pure, while the hydrogen is 99 per cent. Either gas may be purified by passing through red-hot porcelain tubes. TS* FIG. 55. Garuti and Pompili's electrolyzer which combines the small impurity of hydrogen in the oxygen, or of oxygen in the hydrogen, to water which is easily removed. FIG. 56. Garuti and Pompili's electrolyzer This apparatus is made at the Maschinenfabrik Oerlikon. near Zurich, Switzerland. 140 APPLIED ELECTROCHEMISTRY An apparatus in which the separation of the hydrogen and oxygen is effected by a different method is that of Garuti and Pompili. 2 In this cell a partition of iron separates the anode from the cathode, and this partition is prevented from becom- ing an intermediate electrode by keeping the voltage ap- plied to the cell too low for this to take place. The cur- rent flows from the anode to the cathode around the bot- tom of the iron partition. Figure 55 is a longitudinal vertical section through the center, Figure 56 is a horizon- tal section of one end, Figure 57 is a vertical cross section of the apparatus, and Figure 58 a plan view of conductor and electrodes. A tank A of wood lined with iron a contains the electro- lyzer, which consists of an inverted tank A 1 which is di- vided into cells E by longitu- dinal diaphragms. This cell is made of iron and is open 4^ ^ only at the bottom. The l^\ I \ anodes b and cathodes c are placed one in each cell, taking care that each anode is be- FIG. 57. Garuti and Poinpili's electro- lyzer tween two cathodes. The gas passes through an opening at the top of each chamber into the reservoir containing the same gas. The electrodes are insulated from the diaphragms by combs I made of wood, the teeth of which enter the cells and fill the spaces between the electrodes and diaphragm. L is a handle for lifting out the electrolyzer. 2 U. S. Patent 629,070, (1899). THE ELECTROLYSIS OF WATER 141 A 25 per cent solution of potassium hydrate is used. The voltage per cell is not allowed to exceed 3 volts, so there is no danger of the diaphragm acting as an electrode. The diaphragms may be per- forated near the bottom with a large num- ber of small holes, as there is very little danger of the gases becoming mixed at this point. The hydrogen obtained from this ap- paratus is 98.9 per cent pure, the oxygen 97. This apparatus is used in Rome, Tivoli, Brussels, and Lucern. The cell of the Siemens Brothers and Company and Obach 3 employs a parti- tion which consists of metal gauze below the water line. The current is con- ducted through the meshes, which are small enough to prevent the mixture of the gases. Other cells, such as that of Schoop, 4 have nonconducting partitions. These examples complete the different principles on which technical cells for the decomposition of water are built. FIG. 58. Garuti and Pom- pili's electrolyzer 3 Engelhardt, I.e. p. 67. 4 Engelhardt, I.e. p. 44. CHAPTER VIII PRIMARY CELLS A PRIMARY battery is a cell so arranged that the energy of a chemical reaction is obtained as an electric current, and in which the chemicals are not regenerated by passing the current through the cell in the opposite direction. When the battery is run down, fresh chemicals must be supplied. A secondary battery, or accumulator, is a battery in which chemicals are regenerated by passing through the cell, after discharge, a reverse current from some other source. Before the invention of dynamos, primary batteries were the main source of electric energy; but since this method of gener- ating electricity is too expensive for use where a large quan- tity of energy is needed, they were employed only for very light work and for experimental purposes. They are still used extensively for electric bells, for exploding the gases in engines by electric sparks, railroad signals, and similar purposes. Primary batteries of special forms are also the standards of electromotive force, but this is rather a purely scientific branch of the subject than a technical application, and will therefore be omitted. The first primary battery was due to Volta, and consisted in a negative pole of zinc and a positive pole of copper dipping into a solution of salt or dilute acid. The electromotive force of this battery rapidly falls off if an appreciable current is taken from it, on account of the hydrogen liberated on the positive pole. This develops a back electromotive force and also increases the resistance of the cell itself. The battery is then said to be polarized. In order to have a battery that is at 142 PRIMARY CELLS 143 all efficient, polarization must be avoided. In the Smee cell, this was done by substituting platinized silver for the positive pole in place of the copper in the Volta cell. The rough sur- face caused the bubbles of hydrogen to escape more rapidly. In the Grove battery, devised in 1831, 1 the cathode consisted of platinum dipping into nitric acid contained in a porous cup. Outside the cup was dilute sulphuric acid and a zinc negative pole. In this case the nitric acid acts as a depolarizer, oxidiz- ing the hydrogen to water and itself being reduced to nitrous gases. The electromotive force of this battery is between 1.6 and 1.7 volts. The Bunsen cell is a Grove cell with carbon in place of plat- inum for the positive pole. In the chromic acid battery, due to Poggendorff, the electrolyte is a solution of sulphuric acid and potassium bichromate. The positive pole is carbon and the negative zinc, which is withdrawn from the battery when not in use. The chromic acid acts as depolarizer. The electromotive force is about 1.3 volts. These batteries have at present little more than historical inter- est. The use of primary cells is now nearly entirely confined to the Leclanche, the Lalande, and the Daniell cells. Leclanche brought out his cell in 1868. 2 It consists of a zinc rod forming the negative pole and dipping into a solution of ammonium chlo- ride. The positive pole is carbon in contact with manganese dioxide for a depolarizer. When the circuit is closed, zinc goes in solution as zinc chloride and the ammonium radical is deposited on the carbon, which breaks up into ammonia and hydrogen. The ammonia dissolves and the hydrogen is oxidized by the manganese dioxide to water. This depolarization is not rapid, however, consequently not much current can be taken from a Leclanche cell at a time without the voltage dropping consider- ably, but it recovers on standing. The electromotive force of this cell on open circuit differs from one cell to another, varying from 1.05 to 1.8 volts. This cell is put on the market under a large number of different 1 Wiedemann, Die Lehre von der Elektricitat, 1, 867, (1893). 2 Wiedemann, I.e. p. 850. 144 APPLIED ELECTROCHEMISTRY FIG. 59. Carbon of Sampson cell forms and under different names. One of the best Leclanche cells on the market is the Sampson cell. 3 The carbon of this cell is shown upside down in Figure 59. It consists of a fluted hollow cylinder of French carbon pro- vided with a removable seal at the lower end and filled with a mixture of carbon and manganese dioxide. The cell set up is shown in Figure 60. The Lalande cell, brought out in 1883, 4 con- sists of zinc for the negative pole, a 30 per cent solution of potassium hydrate for the electro- lyte, and a plate or box of iron or copper in contact with black copper oxide as depolarizer. The hydrate is protected from the carbonic acid of the air by a layer of oil. The zinc goes in solution as sodium zincate, and the hydrogen deposited on the positive plate is oxidized by the copper oxide. The positive plate may also be an agglom- erate porous plate of copper oxide. The electromotive force of this cell is about 0.9 volt and is very constant. The oxide when reduced to copper is easily oxidized again by heating in the air. The original method of La- lande of making the porous copper oxide plates was to press a moist mixture of oxide, 4 or 5 per cent clay, and 6 to 8 per cent tar, and then to heat to redness. The plates so produced were porous and lasted well. This plate must be reduced to copper over its en- tire surface before its normal rate is reached, on account of the poor conductivity of copper oxide. This is done before assembling the plates. A modern type of the Lalande battery is made by the Edison 8 N. H. Sneider, Modern Primary Batteries, p. 10, (1905). 4 Wiedemann, I.e. p. 854. FIG. 00. The Samson cell PRIMARY CELLS 145 Manufacturing Company at Orange, New Jersey, and is called the Edison-Lalande Battery. This battery, shown in Figure 61, consists of a copper oxide plate between two zinc plates dipping in a 20 or 25 per cent solution of sodium hydroxide. The containing jar is porcelain. The zinc plates have mercury added to them during casting, so that they are amalgamated throughout. The copper oxide ^-^ plates are made from copper scale which is finely ground and then roasted until thoroughly oxidized. The oxidized powder is then moistened with a solution of so- dium hydroxide and pressed into cakes a little larger than desired in the finished product. These cakes are then dried and baked at a bright red temperature, which partially Welds the particles to- FIG. 61. -Edison Lalande battery gether. After cooling, the plates are reduced to copper at the surface by zinc dust, to make them conduct. They are then washed and are ready for use. 5 The hydroxide solution is covered with a heavy mineral oil to prevent its creeping up the zinc plates and corroding them. This battery has an initial electromotive force of 0.95 volt, but on continuous discharge at normal rate it drops to about 0.6 volt. The capacity varies from 100 to 600 ampere hours, depending on the size of the battery. The Daniell cell, brought out in 1836, 6 belongs to a different class of cells, in which there are two liquids separated by a porous partition. The positive pole is copper dipping in a con- centrated solution of copper sulphate, and the negative is zinc dipping in sulphuric acid. Copper is deposited on the positive in place of hydrogen, thus avoiding polarization, and zinc goes in solution forming zinc sulphate. The electromotive force of this cell is about 1.1 volt. The gravity cell, Figure 62, is a form of the Daniell cell 5 Private communication from the company. 6 Wiedemann, I.e. p. 859. 146 APPLIED ELECTROCHEMISTRY FIG. 62. The gravity cell patented by Varley in 1854, but which did not become generally known until 1884. It is now the principal commercial form of the Daniell cell. 7 The gravity cell derives its name from the way in which the two solutions are prevented from mixing. At the bottom of a glass jar is a horizontal copper electrode covered with cop- per sulphate crys- tals and a saturated solution of copper sulphate. On this solution is care- fully poured a di- lute sulphuric acid solution, in which a horizontal zinc electrode is immersed. When in use the migration of the copper ions towards the cathode prevents their reaching the zinc, while if the cell stands on open circuit the copper sulphate would finally reach the zinc by diffusion and cover it with a layer of copper. This cell should therefore al- ways be kept on a closed circuit through a few ohms resistance. Dry cells are a type of primary battery that have recently come into very general use. It is estimated that 50 million a year are manufactured in the United States, a large majority of which are of a standard size, cylindrical in shape, 15 centi- meters long and 6.25 centimeters in diameter. 8 They are essentially Leclanche cells with a very small quantity of elec- trolyte. Their greatest field of usefulness is probably tele- phony and next the ignition through spark coils. 9 The container or outside insulation is usually pasteboard, sometimes waterproofed by paraflfine or pitch. Just inside of the container is the cylindrical zinc negative pole, usually 15 centimeters high, 6.25 in diameter, and 0.3 to 0.55 millimeters thick. Lining the zinc on the inside is a layer of a special grade of pulp board, moistened with a solution of zinc and i Schneider, I.e. p. 54. 8 D. L. Ordway, Trans. Am. Electrochem. Soc. 17, 341, (1910). Burgess and Hambuechen, Trans. Am. Electrochem. Soc. 16, 97, (1909). PRIMARY CELLS 147 ammonium chlorides. The zinc chloride is added for reducing the local action. Inside the pulp board containing the electro- lyte are placed the depolarizer and the positive pole. The de- polarizer is manganese dioxide, mixed with carbon, graphite, or a mixture of both. Graphite is used to give the cell a lower resistance. A carbon rod at the center and surrounded by this mixture is the positive pole. An average composition of this filling mixture is the following : 9 10 parts of manganese dioxide, 10 parts of carbon or graphite, or both, 2 parts of ammonium chloride, 1 part of zinc chloride. Sufficient water is added to give a proper amount of electro- lyte to the cell, depending on the original dryness of the ma- terials, their fineness, the quality of the paper lining, and similar factors. The usual specifications for the manganese dioxide are that it shall contain 85 per cent of the dioxide and less than 1 per cent of iron. The cell is sealed up on top with a pitch composition to hold in the filling material and to prevent the cell from drying. The carbon rod extends above the seal and is provided with a binding screw. The electromotive force of this cell is between 1.5 and 1.6 volts. On a short circuit through an ammeter, a cell will give from 18 to 25 amperes. The energy output of a cell of the dimensions given above, discharged to 0.2 volt continuously, varies from about 20 watt hours when discharged through 2 ohms to 57 watt hours when discharged through 40 ohms. 8 The primary cells described above are comparatively unim- portant compared with one which is not yet realized, but on which a great deal of time and work has been spent. This is the cell in which carbon and oxygen are the elements con- sumed. The present method of producing work by the com- bustion of coal to run steam engines is very inefficient, as only about 15 per cent of this energy is obtained as work, the rest being lost as heat. If it were possible to devise a cell in which carbon and oxygen would unite with the production of an 148 APPLIED ELECTROCHEMISTRY electric current and no other form of energy, at ordinary tem- perature, a much greater amount of energy could be obtained. In order to calculate 10 the free energy, or energy that is obtainable as useful work, of the reaction in question, C + 2 =C0 2 , consider a reaction chamber, as shown in Figure 63, containing carbon, oxygen, carbon monoxide, and carbon dioxide in equi- Cfe CO Ok A"t^Fe(OH) 2 -2F. (4) The sum of these equations is Fe + Ni 2 O 3 - 1.2 H 2 O + 1.8 H 2 O ^ 2 Ni(OH) 2 + Fe(OH) 2 +. (5) This equation represents the final result in the whole cell on discharge, when taken from left to right, and on charge, w r hen taken from right to left. These equations are not reversible in the ordinary sense, however, for they do not show that hydrogen and oxygen are evolved on charging or that the nickelous hydrate is first oxidized to nickel peroxide. The Edison cell is therefore not strictly reversible, and the equations, though written as reversible, are to be taken only as referring to the initial and final states of the cell. It is also to be noticed that in adding the two equations for the iron and the nickel plates the two quantities of electricity, 2 F, cancel out. This means the two quantities neutralize each other, thereby producing the current. The Electrolyte. From the equation (5) it is evident that water is taken up from the electrolyte on discharging by the plates and is given up again on charging. This can be seen by * Foerster, Z. f. Elektroch. 14, 285, (1908). 184 APPLIED ELECTROCHEMISTRY the change in level in the solution on charging and discharging. According to equation (5), 0.9 mole of water would be com- bined or set free to one faraday of electricity passing through the cell. Other experiments made for the purpose of deter- mining this quantity gave an average of 1.45 moles of water. This agreement is not all that could be desired. There is no question, however, that water is removed from the solution on discharging, and it therefore follows that the electromotive force of the battery will decrease with the increasing concentra- tion of the electrolyte. This is verified by the measurements of the following table : 4 NORMALITY OF HYDRATE SOLUTION E. M. F. OF CELL 1.0 1.3510 1.15 1.3368 2.82 1.3377 5.3 1.3349 From what has preceded, it will be evident that the current efficiency and capacity depend on each other. If the battery is not fully charged, the current efficiency will be high, but the full capacity is not obtained. This can be obtained only by charging after gas evolution has begun, which reduces the current efficiency. When the cell was charged and discharged at the normal rate of 4 hours, the ampere hour efficiency was about 75 per cent, and the voltage efficiency about 70 per cent, making the energy efficiency about 50 per cent. 6 5 Kennelly and Whiting, Trans. Am. Electrochem. Soc. 6, 146, (1904). CHAPTER XI THE ELECTRIC FURNACE 1. GENERAL DISCUSSION THE electric furnace industries are at present in a state of rapid development. This is due partly to the manufacture of a large number of new products made possible by the high temperature attainable in the electric furnace, and partly to improved methods in the manufacture of products previously obtained by other methods. The electric furnace was probably first used on a commercial scale by the Cowles Brothers in 1884 in their manufacture of aluminum alloys, but the rapid increase in its use began about 1893 with the production of calcium carbide, carborundum, and aluminum. In the manufacture of many electric furnace products, heat at a high temperature is the form of energy that brings about the change desired. The question naturally arises, how is it possible that it should be economical to obtain heat from such an expensive form of energy as electricity. There are several reasons why it is economical. In the first place, the temper- ature required for the formation of many electric furnace prod- ucts is above that attainable by any commercial fuel. In such cases it is evident that if the product is to be formed at all, it must be formed in an electric furnace. On the other hand, it has been found economical to use heat generated from electricity in cases where fuel was formerly used. This is due to a simplifi- cation in the apparatus and a saving of time and labor. While electric heat costs more per unit, it may be possible to reduce the time during which it has to be applied to such an extent 185 186 APPLIED ELECTROCHEMISTRY that the quantity of heat required is so much less than when fuel is used that it more than saves the extra cost per unit. This is often the case on account of the fact that electric heat is generated inside the furnace or container just where it is wanted, while in the use of fuel the heat is generated outside the furnace and has to penetrate the walls before reaching the material to be heated. It is evident that more heat will be lost in the latter than in the former process. In those furnaces in which the electricity flows through a core especially made for the purpose and not through the charge itself, the temperature to which the core is raised is one of the factors that determines the time required to bring the charge up to the desired temperature, since the flow of heat between two bodies is proportional to their difference in tem- perature. Furnaces may be divided into three classes : arc furnaces, resistance furnaces, and induction furnaces. In the first, as the name indicates, the source of heat is an arc. A solid body to be heated is placed near the arc and is heated by radiation. By adjusting this distance the temperature to which it is raised may be regulated. In case a gas is to be heated, the passage of the arc through the gas itself brings about the desired result. In the resistance furnaces the current generates heat by passing some suitable resistor. It is evident that arc furnaces are simply resistance furnaces where the resistor is a gas; but nevertheless this distinction is a convenient one. Resistance furnaces may be of two kinds, first, those in which the current passes through the charge to be treated and develops heat in consequence of the resistance of the charge, and second, those in which the current passes a resistor surrounded by the charge. The latter furnace is used in those cases where the charge itself does not conduct well. The first class of resistance furnace may be divided into two classes, in which (1) the thermal effect is alone active, and (2) in which electrolysis also takes place. The induction furnace is the latest type, and is used in the steel industry. The metal to be heated forms the secondary winding of a transformer, and forms a closed ring in an annular THE ELECTRIC FURNACE 187 crucible. A current is induced from the primary winding sufficiently great to melt the metal. The following table summarizes this classification. ELECTRIC FURNACES 1. 2. 3. Arc Resistance Induction 1. The charge con- 2. Current conducted ducts the current. by a special resistor. 1. Withelec- 2. Without elec- trolysis, trolysis. 2. ELECTRIC FURNACE DESIGN In spite of the fact that the heat is generated inside the furnace, there is always some heat lost by conduction through the walls of the furnace, through the electrodes, and in some cases by hot gases. To increase the economy of furnaces these losses must be made as small as possible. The case when the loss is due to gases requires no special consideration, but it will be desirable to consider the losses through the walls and the electrodes. If H equals the number. of calories conducted in one second through a wall of cross section S, thickness I, and specific con- ductivity &, when the difference in temperature of the two faces is T and no heat is lost through the ends of the walls, rr SkT then H- l In the case of a furnace, the cross section of the wall is not constant, but increases from the inner to the outer surface. Generally in making this calculation the average cross section is taken. Where the walls are thin, this is fairly accurate, but with thick walls a very great error may be introduced. 1 1 Carl Hering, Trans. Am. Electrochem. Soc. 14, 215, (1908). The discussion in the text is taken from this article. 188 APPLIED ELECTROCHEMISTRY For a complete sphere, inner surface s, outer surface S, and thickness of wall I, the heat conducted per second for unit dif- ference of temperature is 2 TT _ where D is the outside and d the inside diameter. For a cube 6 TcDd I I where D is the length of the outer edge and d that of the inner edge. For a cylindrical shell of length (7, thickness of wall Z, outside diameter 2>, and inside diameter c?, 2-3 log" 2 The derivations of this and the following formulae, not given by Bering in the article referred to, are very simple. The resistance of a spherical shell of thickness dx, where the radius of the shell is x, is dE = , if r specific resistance. 4 7TX 2 Integrating between the limits x = ai and x 2 , where a\ and 2 are the inner and outer radii respectively, T-J T f Ct2 ~T^\~a^L But if 8 is the outer surface and s the inner, #=4 7ra 2 2 , s=4 IT i 2 , and 2 i = the thickness of the shell. Substituting these values, To get the formula for the cylinder of length O all that is necessary is to integrate the equation dE= between x = a 2 and x = ai, 2 TTCX giving JR^-^- log, 2S. ^ TTO (Zl For the cubical frustum dB = r** , whence 7?= - r f ^A = -iL = H, wx 2 n V ai2 / Vs^ ^-D where n is given by the equation S= THE ELECTRIC FURNACE 189 The curves in Figure 78 give an idea of the error that would result from using the mean value of the cross section in place of the above formulae. As abscissae are taken the thickness of wall in terms of the inner diameter or edge, and as ordinates the conductivity for one degree difference in temperature and for a substance whose specific conductivity is one. The dotted FIG. 78. Heat loss as function of thickness of walls lines show the conductivity as given by the approximate formula, and the full lines show the true value. It is evident that the greatest errors occur in the cases of the cube and sphere, where they are quite appreciable when the thickness of the wall equals one half the diameter or one half the inner edge. In Table 20 the values of the heat conducted through the walls of the three typical furnaces are collected, which are those given 190 APPLIED ELECTROCHEMISTRY in the plot. 3 The conductivity and difference in temperature are assumed unity. TABLE 20 Heat Conductivity of Spherical, Cubical, and Cylindrical Furnaces THICKNESS SPHERES CUBES CYLINDERS CONDUCTANCE CONDUCT ANCE CONDUCTANCE By Correct Formula By Approxi- mate Formula By Correct Formula By Approxi- mate Formula By Correct Formula By Approxi- mate Formula 0.10 37.70 38.3 72.0 73.2 34.40 34.60 0.15 27.20 28.2 52.0 53.8 23.90 24.10 0.20 22.00 23.2 42.0 44.4 18.70 18.80 0.25 18.90 20.4 36.0 39.0 15.50 15.70 0.30 16.80 18.6 32.0 35.6 13.40 13.60' 0.35 15.30 17.5 29.1 33.3 11.80 12.10 0.40 14.10 16.7 27.0 31.8 10.70 11.00 0.45 13.30 16.1 25.3 30.7 9.79 10.10 0.50 12.60 15.7 24.0 30.0 9.06 9.43 0.60 11.50 15.3 22.0 29.2 7.97 8.38 0.70 10.80 15.2 20.6 29.0 7.18 7.63 0.80 10.20 15.2 19.5 29.1 6.58 7.07 0.90 9.77 15.4 18.7 29.5 6.10 6.63 1.00 9.42 15.7 18.0 30.0 5.72 6.28 1.50 8.38 17.8 16.0 34.0 4.53 5.24 2.00 7.85 20.4 15.0 39.0 3.90 4.71 3.00 7.33 26.2 14.0 50.0 3.23 4.19 The following example will show how this table may be used in the case of a furnace of one of these types. Let the inner diameter of a spherical furnace be 15 inches, the thickness of wall 9 inches ; to find the heat conductance if the wall consists of infusorial earth whose specific heat conductivity is "k = 0.001 in gram calorie cubic inch units, and if the difference in tem- perature between the inside and outside face is 700 C. The 8 Hering, I. c. In the original table four and five places of significant figures are given. Since the specific conductivity of refractory substances at high tem- peratures is not known to more than two places, only three places are here retained. THE ELECTRIC FURNACE 191 thickness in terms of the diameter is -= = 0.6. Opposite d 15 0.6 in the table the conductance is 11.5. This number evi- dently must be multiplied by c?, &, and 700, giving a loss of 121 grams calorie per second. On the other hand, if the loss is given and the temperature difference and conductivity are known, the corresponding thickness can be found. In the case of the cylinder, the conductivity calculated is for the cylindrical part alone. These values must therefore be multiplied by the length of the cylinder, but not by the inside diameter, and the loss at the two ends must be added. Unfortunately heat conductivities of refractory substances are not accurately known above 1000 C. Recently, however, the mean conductivities between room temperature and 1000 C. of a number of refractory substances have been determined un- der the direction of Le Chatelier by Wologdine. The results 4 have been collected by Queneau in Table 21. Data are also TABLE 21 Conductivity of Refractory Materials MATERIAL CONDUCTIVITY Gram Calorie per Cm. Cube per 1 C. Diflf. in Temp. Relative Conductivity in Per Cent of Value for Graphite 0.0250 0.0231 0.0071 0.0057 0.0042 0.0039 0.0038 0.0035 0.0033 0.0027 0.0023 0.0020 0.0018 100.0 92.0 28.0 23.0 17.0 16.0 15.0 14.0 13.0 12.0 9.3 7.8 7.1 Carborundum brick .... Fire brick Checker brick Silica brick .... Infusorial earth brick . . . Electrochem. and Met. Ind. 7, 383, (1909). 192 APPLIED ELECTROCHEMISTRY given in the same article on the porosity and gas permeability of these materials. The principal refractory substances for electric furnaces are" carbon, carborundum, and siloxicon. 5 The use of siloxicon is limited to temperatures below that at which it is converted into carborundum, and of carborundum to temperatures below which it breaks up into silicon and graphite. These sub- stances all have a higher thermal conductivity than the other less refractory materials, as seen in the above table, and for this reason it is usual to build furnace walls in sections, with highly refractory material inside, where the temperature is highest, and with material offering a high resistance to the passage of heat outside. Carborundum, for instance, is one of the most refractory materials, but as seen from the table its conductivity is high. It would, therefore, be well to use this as a lining of such a thickness that the temperature on the outside of the lining would not be too high for some material with a lower heat conductivity, such as fire brick or infusorial earth. Knowing the dimensions, the total loss in power, and the conductivity, the temperature of the cool side of the lining is easily calculated. 6 The loss of heat due to conductance through the electrodes will next be considered. This loss is made up of two quanti- ties, the heat generated in the electrode by the passage of the current and the heat which would flow from the hot to the cold end if the temperature at the hot end were maintained without passing a current through the electrode. The following dem- onstration 7 will show how the total heat loss due to the elec- trodes is related to these two losses, and how electrodes should be proportioned to make this loss a minimum. In Figure 79, let ab be a conductor of heat and electricity imbedded, except at its ends, in a perfect insulator of heat and electricity. Let the temperature at a be T C. and at 6, C. 5 FitzGerald, Electrochem. and Met. Ind. 2, 349, (1904). 6 For examples see Hering, Electrochem. and Met. Ind. 7, 11, (1909). 7 Hering, Trans. Am. Electrochem. Soc. 16, 287, (1909) ; also Electrochem. and Met. Ind. 7, 442, (1909). a THE ELECTRIC FURNACE 193 Let a current also ^ JH pass through the ^ K?TT*T"T!T^f?.T^ ; * electrode. The ^ problem is to find the quantity of FlQ 79 heat flowing out the cold end, when a steady state has been reached. Let X total energy in watts pressing out of the cold end. x = energy passing any cross section at distance I from the hot end. H= number of watts that would flow from the hot to the cold end were there no current. h number of watts entering the hot end. W= number of watts generated by the current in the electrode. W w = where L = total length of electrode. L T= total fall in temperature from hot to cold end, when cold end is at 0. t = temperature at any length I from hot end. L = total length in centimeters. I = any distance from hot end. S = cross section in square centimeters. k = mean heat conductivity for the given range of temper- ature in gram-calorie centimeter centigrade degree units. r = mean electrical resistivity for the given range in ohms for a cube of one centimeter edge. 1= current in amperes. R = total resistance. j is the factor 4.19 by which a given number of calories per second is multiplied to change to watts. Let dl be an infinitely short section at distance I from the hot end, and let the heat flowing into this section be x. The heat generated in the section by the current will be wdl=dx. (1) 194 APPLIED ELECTROCHEMISTRY Also .-/feS (2) dl where is the heat gradient at I. dl Differentiating this gives dl, (3) and eliminating dx between equations (3) and (1) d?t = w_ Pr dl* jkS JkS* Since r and k are functions of , to be strictly accurate these quantities should be expressed as such before integrating. For the sake of simplicity, however, mean values for r and k for the temperature interval considered are taken, and these quantities in equation (4) are treated as constants. Integrating once under this assumption gives dt Wl TJ = a ~~ "v"? ' dl jkS and a second time In this equation a and b are determined by the fact that when Z = 0, t=T, and when l=^t = Q. Substituting these values in (6) gives Substituting this value of a in (5) and the value of thus ob- dl tained in (2) gives This equation states that the energy passing any given cross section is equal to the energy that would pass were no current flowing, minus one half the PR energy, plus the PR energy generated in the hot end. When L = l, since wL = W, THE ELECTRIC FURNACE 195 This states that the energy passing out the cold end as heat equals the energy that would pass out when no current is flow- ing, plus one half the I^R energy. W Suppose that in (7) I = 0, then x h and h = If -- - . In order that no heat shall enter the hot end, h = 0, whence W ff= . The last equation states that if no heat enters the hot 2 end from the furnace, the heat flowing from the hot to the cold end of the electrode if there were no current equals ^ I*R. Now the product of H x - =? - , which is independent of S 2 2 and L. When the product of two variables is a constant, their sum is a minimum when the two variables are equal; that is, in TTT -prr the equation JT= H-\ -- , X will be a minimum when H= -> 2 2 or the minimum loss = I^R. Substituting the values of H and TF W in H = , we have the equation 2 Solving this for L y = 0.346jfy-^-, (9) and substituting this value in the equation (8), If, in place of using mean values of the specific heat conductiv- ity and electrical specific resistance, the variable values 8 k t = & (1 + at) and r t = r (l + a^) are substituted in the formulse above, the following results are obtained : 8 H. C. Richards, Trans. Am. Electrochem. Soc. 16, 304, (1909). 196 APPLIED ELECTROCHEMISTRY = 0.346 5 a \ ~ 3 a 12 T\ / and The errors introduced by using mean values of k and r and treating them as constants will be small unless the temperature coefficients are enormous. As was shown above, the minimum loss of one electrode is J?R or le. Substituting this in (10), e = 2.89V&r2 r . 0-1) This voltage is seen to be dependent only on the thermal con- ductivity, electrical re- sistivity, and tempera- ture difference of the ends of the electrodes, which means that for every material there is a characteristic mini- mum drop of potential in the electrodes for one degree difference in temperature below * which it is not possible to go without increas- ing the loss. This minimum drop in potential has been called the electrode voltage. The temperature distribution in the electrode is given by the equation : t^T-^^^L-.-^ n^ T * Ct .'7_* Ct '1. O' FIG. 80 obtained from equation (6) by substituting in the values of a and b. The variables being t and Z, the curve is evidently a parabola. If no current flows, w = and the equation becomes the straight line ceb in Figure 80. THE ELECTRIC FURNACE 197 Making T=Q gives t = o"Vy ' ^e P ara bola p. To find the temperature distribution for minimum loss, solve for T in '. O -TTT the equation JKT = , and substitute in (12), obtaining the parabola P. When - is greater or less than H, the tem- perature distribution is given by P f or P" respectively. In any problem involving the design of electrodes, the tem- perature difference between the hot and cold ends of the elec- trode and the kilowatts to be absorbed in the furnace will be given. From the value of the power the voltage would then be made as high and the current as low as practicable. From formula (9) compute the proportion of the section to the length. The length, which should be as short as possible, will be determined by the thickness of the walls of the fur- nace. Having fixed the length, the section is then obtained from the ratio of the section to the length. The two remain- ing factors which must be known are the values of the heat and electrical conductivities of carbon and graphite, the only two substances used for electrodes in resistance furnaces. These values have not yet been determined accurately for high temperatures, but the mean values have been determined by Bering between 100 C. and 900 C. 9 The method of deter- mining heat conductivity depends on the demonstration above. If in equation (13) Z=0, then t= ^and k= In order to measure &, a conducting rod of length L and sec- tion S, embedded in a nonconducting material, is heated by a measured amount of electrical energy and the temperature T measured at the center. In order to have no heat pass out the sides of the rod, it is surrounded by a number of similar rods at the same temperature as the one measured. The electrical conductivity is obtained from the ammeter and voltmeter read- 9 Trans. Am. Electrochem. Soc. 16, 317, and 315, (1909). 198 APPLIED ELECTROCHEMISTRY ings and the dimensions. The values in Table 22 have been obtained by this method. 9 The units are centimeters, gram calories, and ohms, and centigrade degrees. TABLE 22 GKAPHITE CARBON BETWEEN 100 C. AND TEMPERATURE GIVEN Heat Electrical Heat Electrical Conductivity Kesistivity Conductivity Eesistivity 900 0.291 0.000820 0.129 0.00276 390 0.339 0.000838 360 0.0890 0.00422 The accuracy of these figures is estimated at a few per cent. The electrode voltage (equation (11)) from these data for one degree for graphite is 0.0447 and for carbon is 0.0639, which means that the minimum loss for carbon is about 50 per cent greater than for graphite. Later measurements by Hering 10 gave results from which the following Table 23 has been com- puted. The values of heat conductivity and for electrical resistivity are for centimeter cubes. TABLE 23 TEMPERATURE, C. Hot End Cold. End Carbon 300 40 0.0891 0.00422 701 50 0.124 0.00381 902 60 0.130 0.00377 Graphite 355 66 0.339 0.000837 516 70 0.325 0.000827 707 87 0.309 0.000802 10 Trans. Am. Electrochem. Soc. 17, 166, (1910). THE ELECTRIC FURNACE 199 The following data were obtained by Hansen. 11 The units are the same as in the table above. TABLE 24 ACHESON GRAPHITE NATIONAL CARBON Co.'s ELECTRODES TEMPERATURE Heat Con- Electrical Heat Con- Electrical ductivity Kesistivity ductivity Resistivity 25 0.00066 to 0.00287 to 0.00260 0.0254 Between 3200 and 200 0.00081 2830 and 30 0.155 3500 and 30 0.0155 The electrical and thermal conductivities of carbon elec- trodes cannot be determined above 1600, because on cooling the values do not come back to the original ones, due to a par- tial conversion of the carbon into graphite. 11 Besides the loss in the electrode itself, a large loss occurs at the contact between the electrode and the cable, due to the contact resistance. This resistance varies with the current density, and where brass clamps are used on graphite it amounts to 0.0117, 0.0045, and 0.0039 ohms per square centi- meter for current densities of 3.7, 5.6, and 7.4 amperes per square centimeter. 11 With the aid of the constants above, a numerical example may be given. Let the capacity of the furnace be 500 kilo- watts, the current 10,000 amperes, and the temperature 1700 C. inside and 100 at the cold end of the graphite electrode. Assuming r = 0.000820 and k = 0.291, by formula (10), X= 17.8 kilowatts for each electrode. Assuming for carbon, r = 0.00276 and Jc = 0.129, X= 21.8 kilowatts. Of course, the cross sections of the graphite and carbon electrodes are not equal for equal lengths. 11 Trans. Am. Electrochem. Soc. 16, 329, (1909). 200 APPLIED ELECTROCHEMISTRY The discussion so far has been for the case that the dimen- sions of the furnace and the power to be applied in order to bring about a desired result are known. If these are not known, an experiment would usually be made on a small scale in order to determine the relation between the size of fur- nace and the power. There are two cases to be considered, (1) when there is a central core for carrying the current, and (2) when the charge to be heated itself carries the current. In the first case the heat has to be conducted from the core to the surrounding charge. 12 The rate of this flow is propor- tional to the difference in temperature of the core and the sur- rounding charge, the thermal conductivity of the charge, and the surface area of the core. If heat is generated in the core at a given rate, the temperature to which it will rise in a given time will depend on the specific heat of the core and the rate at which the heat flows into the surrounding charge. This rate of flow depends on the area of the core and the conduc- tivity of the charge. Suppose that to bring about the desired reaction in a given charge with a core of a given material ex- periments are made with a small furnace until the conditions are found under which the desired reaction is brought about. This means that a definite amount of heat must pass per unit surface of the core, which is a constant for these materials and is independent of the dimensions. If the voltage is E and the TFT current J, the energy in watts per unit surface is a = - > 2 irJrL when P is the radius and L the length of the core. Collecting the constants in one factor, this may be written PL = AEI. If r is the specific resistance of the core, we also have XT T T = = B . For any furnace of any other dimensions L v and P v the voltage and current E l and I are given by the XT T equations P^ = AE^ and -1=1? 1. From these equa- ~l *! tions we could solve for the new values E and I v if L 1 and P 1 M FitzGerald, Electroc-hem. and Met. Ind. 2, 342, (1904). THE ELECTRIC FURNACE 201 are given. Usually, however, the power is given, and the proper dimensions L^ and P l are desired. Solving for these quantities, P 1 = and , A The following is an example of the use of these formulae. It was desired to design a 200-kilowatt furnace using a current of 4000 amperes and 50 volts. Experiments on a small scale showed that the right conditions were obtained with 200 am- peres at 100 volts and a core 365 centimeters long and 5.1 cen- timeters in radius. From these values the proper length and radius for the large furnace are found to be 495 centimeters and 37.6, respectively. For the second case, where the current passes through the charge itself, it is simply necessary to know the amount of heat required to raise a given mass to the desired temperature, that is, the number of watts per unit mass. If the specific heat of the charge is known, this can be computed ; if not, an experiment on a small scale with a given mass will determine the energy required. CHAPTER XII PRODUCTS OF THE RESISTANCE AND ARC FURNACE 1. CALCIUM CAKBIDE THE discovery of calcium carbide is due to Wohler, 1 who prepared it by the action of carbon on an alloy of calcium and zinc. Even previous to Wohler, E. Davy had also produced it in an impure state without identifying it. 2 The commercial importance of calcium carbide, however, dates from its redis- covery by Thomas L. Willson, 3 which was nearly simultaneous with that of Moissan (1892). The reaction between lime and carbon by which calcium carbide is produced is the following : As indicated, this is a reversible reaction, and according to the Phase Rule has one degree of freedom ; that is to say, at a given temperature there is one definite pressure of carbon monoxide which corresponds to equilibrium. At 1475 C. this pressure has been found to be 0.82 millimeter of mercury. 4 Above 1500 calcium carbide decomposes into its elements, but of course not as rapidly as it is produced, otherwise its manufacture would be impossible. When calcium carbide is formed from calcium and diamond, 7250 calories are absorbed at room temperature. When formed from lime and carbon, 121,000 calories are absorbed at room temperature, and the temperature coefficient of the heat of the 1 Ann. d. Chem, und Pharm. 125, 120, (1863). 2 Lieb. Ann. 23, 144, (1836). See Abegg, Handbuchder anorganischen Chem. 2, 119. 8 Lewes, Acetylene, p. 24, (1900). 4 Thompson, Proc. Am. Acad. 45, 431, (1910) ; also Met. and Chem. Eng. 8, 327, (1910). 202 PRODUCTS OF THE RESISTANCE AND ARC FURNACE 203 reaction has been calculated to be 3.3 calories per degree. 6 The fact that heat is absorbed when the above reaction proceeds from left to right shows that the equilibrium pressure of carbon monoxide increases with the temperature, and it can be calcu- lated that at about 1840 the pressure equals one third of an atmosphere. If carbon were heated in the presence of air much above red heat, all the oxygen would be converted to carbon monoxide, and if none escaped, its resulting partial pressure would be one third of an atmosphere. It would therefore be necessary to heat carbon and lime to a temperature above 1840 C. before carbide could be formed. In actual practice, however, the partial pressure of carbon monoxide would be less than one third of an atmosphere, in which case carbide could be formed at a lower temperature. Taking these facts into consideration, it does not seem probable that 2000 C. is exceeded in actual practice, for high temperature would accelerate the decomposi- tion of the carbide already formed. This explains the fact that a resistance furnace, in which the temperature is lower than in the arc, gives better yields than an arc furnace. 6 Commercial calcium carbide is dark colored and crystalline but if pure it is colorless and transparent. 7 It has a density at 18 of 2.22, and is insoluble in all known solvents. It is a powerful reducing agent. If heated with metallic oxides it gives, according to circumstances, an alloy of the metal in question with calcium or the metal itself, probably according to the reaction. 7 3 M 2 + CaC 2 = CaO + 3 M 2 + 2 CO or 5 M 2 O + CaC 2 = CaO + 5 M 2 + 2 CO 2 . It further has the property of absorbing nitrogen according to the equation CaC 2 + N 2 =CaCN 2 + C, forming calcium cyanamide. This is an important method of fixing atmospheric nitrogen, and will be referred to later under that heading. 6 Thompson, Trans. Am. Electrochem. Soc. 16, 202, (1909). 6 Tucker, Alexander, and Hudson, Trans. Am. Electrochem. Soc. 15, 411, (1909). 7 Abegg, Handbuch der anorganischen Chem. 2, p. 121. 204 APPLIED ELECTROCHEMISTRY The principal use of calcium carbide is to produce acetylene for illumination. This gas is evolved when the carbide is treated with water, according to the reaction: CaC H 2 CaO C 2 H 2 . The first to produce calcium carbide on a commercial scale, as stated above, was Thomas L. Willson, at the Willson Aluminum Works at Spray, 8 North Carolina. Willson was attempting to reduce lime by heating with carbon, hoping to get calcium with which to try the reduction of alumina. It was by accident that the material produced was found to react with water and give off an inflammable gas. Soon after this discovery Willson's plant at Spray was investigated by Houston, Ken- nelly, and Kennicutt. 9 Two runs were made with the purpose of determin- ing the cost of manufac- turing calcium carbide under conditions existing at that place. There were two furnaces built in one structure, as shown in Figure 81, the walls and partition of which were brick, while the front was only partly cov- ered by cast-iron doors. The floor space of each furnace was 3 by 2J feet. The furnaces united at a height of 8 feet into a single chimney for carry- ing off the gases. The base of the furnaces con- FIG. 81. Carbide furnace at Spray, North Carolina 8J. W. Richards, Electrochein. Ind. 1, 22, (1902). The date given by Richards is 1891. This is evidently too early ; see note 3. 9 Progressive Age, 14, 173, (April 15, 1896). (Published at 280 Broadway, New York City.) PRODUCTS OF THE RESISTANCE AND ARC FURNACE 205 sisted of a heavy piece of iron between, 1 and 2 inches in thick- ness, 6 feet in length, and 2| feet in width. The iron plate was completely covered by two carbon plates between 6 and 8 inches thick. These formed the lower electrode. The upper electrode of each furnace was a carbon block 12 by 8 inches in section and 36 inches long, protected by an iron casting -fa inch thick. The space between the casting and carbon was filled with a mixture of hot pulverized coke and pitch. The first run lasted 3 hours with an aver- age activity supplied to the furnace of 144 kilowatts at approximately 100 volts. FlG - 82. - Longitudinal ver- . tical section of the first car- ThlS made a total power consumption of tide furnaces at Niagara 432 kilowatt hours, yielding 98.0 kilo- Falls grams of 79 per cent pure carbide. The second run lasted 2 hours and 40 minutes with an average activity of 146.7 kilowatts, making the ^ / total power consumption 388.5 killowatt hours and yielding 87.5 kilograms of 84 per cent carbide. This is about 0.225 kilograms of carbide per kilowatt hour. The cost of producing carbide at Spray, working the furnaces 365 days a year and 24 hours a day, was estimated at about $33 per 2000 pounds of impure carbide. This estimate, however, is made up of a large number of items that would be considerably changed for other places. The largest producer of carbide in the FIG. 83. Transverse ver- United States is the Union Carbide Com- S35?i; p^y. whose W01 ' ks are at Nia s ara Falls -' ara Falls Their first furnaces were of the Willson type, in which the lower electrode was a small car which could be removed, when filled with an ingot of carbide, to make room \ LJ Air 206 APPLIED ELECTROCHEMISTRY for another, as shown in Figures 82 and 83. This type has been displaced at Niagara Falls by the Horry rotary continuous fur- nace, introduced in 1898 and shown in Fig- ure 84. 10 It consists of an iron wheel 8 feet in diameter and 3 feet in width, with an an- nular-shaped space around the circumfer- ence in which the car- bide is formed. The electrodes project ver- tically down into this space. Lime and car- bon are fed in, and as carbide forms, it is removed from the electrodes by the rotation of the furnace. Iron plates hold the carbide in place while under the influence of the current. When the rotation has carried it to the other side of the furnace, it has had time to cool, as there is only one complete rotation a day. The outer plates are then removed, and the carbide is broken off in pieces 6 to 9 inches thick. Each furnace takes 3500 amperes at 110 volts and produces 2 short tons of carbide a day. FIG. 84. Horry carbide furnace COUNTRY PRODUCT IN METRIC TONS 38,000 Italy 32000 27000 Norway 25000 Switzerland 20000 Austro- Hungary 12,000 Sweden .... . . . . 12000 9,000 10 Lewes, ibid. p. 207 ; Richards, Electrochem. Ind. 1, 22, (1902) ; Haber, Z. f. Elektroch. 9, 834, (1903). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 207 The production of the Union Carbide Company from year to year has not been made known. The preceding table shows the estimated output of the world for 1908. u In 1902 the Union Carbide Company sold carbide to home consumers at about $70 a ton, but exported it for $50 a ton. 12 In 1907 the price was still $ 70 a ton in this country. In Europe 13 the form of furnace still used is of the Willson type. In some cases the ingot is formed on a truck that can be removed when full, and in others a stationary crucible is used. In the former case it has been found an improvement to have two electrodes suspended over the truck, so that the truck is no longer in the electric circuit. In the case of fixed crucibles the capacity has been increased in some cases up to 6000 kilowatts, and a more satisfactory method of tapping has been devised. Formerly the solid carbide formed around the tap hole had to be broken away, but the later method consists in inserting an iron rod connected to the upper electrode into the tap hole, where an arc is formed between the rod and the solid carbide. The iron and carbide are both melted by the arc, and an opening is formed through which the melted carbide can flow out. With regard to the power required for the production of car- 'bide, the only figures of any practical importance are not those obtained by calculation, but those obtained in actual practice. The original plant of Willson produced 5.4 kilos per kilowatt day of 24 hours 9 of 80 to 85 per cent carbide. At Meran the yield is 5.8 kilos of 78 per cent carbide per kilowatt day. 14 At Foyers in Scotland the yield per kilowatt day of 24 hours is 4.2 kilos of 87 per cent carbide. 15 At Odda, Norway, it lies between 4.5 and 5.2 kilograms. 16 The materials 17 used in making carbide are freshly burnt lime and carbon in the form of anthracite coal, metallurgical coke, or charcoal. Ordinary gas coke has too many impurities for this purpose. Charcoal is used only where one of the other forms of carbon cannot be obtained, as it generally contains u Min. Ind. 17, 100, (1908). 12 Min. Ind. 11, 76, (1902). 13 See Conrad, Electrochem. and Met. Ind. 6, 397, (1908). 14 Lewes, Acetylene, p. 242. 16 Lewes, I.e. p. 262. is Electrochem. and Met. Ind. 7, 213, (1909). 17 Lewes, pp. 264-284. 208 APPLIED ELECTROCHEMISTRY considerable traces of phosphates, which appear in the acetylene generated from the carbide in the form of phosphureted hydro- gen. The reaction requires 36 parts of carbon to 56 of lime. In most ingot carbide furnaces 100 parts of lime to 70 of car- bon are used. In furnaces from which the carbide is drawn off in the liquid state a higher proportion of lime is used in order to lower the melting point of the carbide. This, of course, has the result of making the carbide less pure. It was at first supposed that fine grinding of the materials was necessary, but it has since been found that pieces as much as one inch in diameter may be used. 18 2. CARBORUNDUM Carborundum is the trade name for the carbide of silicon, which has the formula CSi. It was probably first produced by Despretz in connection with experiments on refractory ma- terials, 1 in the course of which he heated a carbon rod em- bedded in sand by passing an electric current through the rod. He obtained a very hard tube of six times the diameter of the carbon rod, lined on the inside with quartz in the form of lampblack. It seems probable that in this experiment some carborundum was formed, though no mention is made of crystals. It seems more certain that carborundum crystals were obtained by R. Sidney Marsden, 2 by heating for several hours silver or an alloy of silver and platinum in a Berlin porcelain crucible with amorphous carbon considerably above the melting point of silver and then cooling slowly for 12 to 14 hours. On dissolving the silver in nitric acid it yielded from its interior a number of beautiful crystals of the hexagonal system and varying in color from light yellow to dark brown, or even black. Other crystals were found in the form of hexagonal prisms, but these were in most cases colorless and transparent. The colored crystals were doubtless crystallized carborundum, formed from the silica glaze on the crucibles and the amorphous carbon. 18 Blount, Practical Electrochemistry, p. 230, (1901). 1 C. R. 89, 720, (1849). 2 Proc. Royal Soc. of Edinburgh, 11, 37, (1880-1881). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 209 The white crystals were evidently silica, as they dissolved when boiled in hydrofluoric acid. In 1886 A. H. Cowles 3 obtained some hexagonal crystals from his furnace on attempting to melt quartz. This was an- alyzed and thought to be a suboxide of silicon. On seeing Ache- son's Carborundum at the Chicago Exposition in 1893, Cowles recognized its similarity with his so-called suboxide of silicon. This resulted in a lawsuit between the Cowles Electric Smelting and Refining Company and the Carborundum Company. 4 Schut- zenberger and Colson hacLsuspected the existence of a com- pound of the formula Si 2 C 2 as early as 1881, 5 and in 1892 Schiitzenberger 6 obtained the amorphous carbide of silicon by heating together silicon, silica, and carbon, and determined its composition. Its color was a clear green. Finally, Moissan 7 has made crystallized carbide of silicon in the following different ways : 1. Carbon was dissolved in melted silicon between 1200 C. and 1400 C. from which crystals of carbide several milli- meters long were obtained by dissolving the silicon in a boiling mixture of concentrated nitric acid and hydrofluoric acid. 2. By heating silicon and carbon in the proportion of 12 parts of carbon to 28 parts of silicon. The mass of crystals obtained was easily purified by first boiling in a mixture of concentrated nitric acid and hydrofluoric acid and by then treating with nitric acid and potassium chlorate. The crystals were frequently colored yellow, but could be obtained completely transparent. 3. By heating a mixture of iron, silicon, and carbon in the electric furnace, giv- ing a metallic fusion containing crystals of carbide of silicon. The excess of iron or silicon was then dissolved. 4. By heating silica and carbon in the electric furnace. 5. By the action of the vapor of silicon on the vapor of carbon. This experiment was made in a small carbon crucible containing fused silicon. The bottom of the crucible was heated to " the highest tempera- 8 Proc. of the Soc. of Arts for 1885-1886, p. 74, Boston. 4 FitzGerald, Carborundum, in the Engelhardt Mongraphien iiber Ange- wandte Elektrochemie. *C. R. 92, 1508, (1881). 6 C. R. 113, 1089, (1892). 7 Moissan, The Electric Furnace, translated by Lehner, p. 274, (1904). 210 APPLIED ELECTROCHEMISTRY ture of the electric furnace." After the experiment, slightly colored, very hard and brittle crystals in prismatic needles of carbon silicide were found. The description of this experi- ment is far from convincing. If the crystals were found in the silicon, there is no evidence of the action of one vapor on the other, but even the original article 8 does not state where the crystals were found, which would be necessary to decide the question. In 1891 at Monongahela, Pennsylvania, E. G. Acheson 9 dis- covered the crystallized carbide of silicon, in carrying out some experiments with the object of producing crystallized carbon. The object was to dissolve carbon in melted silicate of alu- minum, or clay, and by cooling to cause the carbon to crystallize. The first experiments were carried out in an iron bowl lined with carbon in which was placed a mixture of carbon and clay. The mixture was fused by means of an electric current passing between the bowl and a carbon rod directly over it. On fusion a violent reaction took place, and after cooling a few bright blue hard crystals were found. These were first supposed to be carbon, but later were taken for a compound of alumina or corundum and carbon, from which the name carborundum was made up. Subsequent to this it was found that better results were obtained when silica was used in place of clay, and when common sodium chloride was added. The reason for this was evident when the following analysis of the product was made: Silicon 62.70 per cent Carbon 36.26 per cent Aluminum oxide and ferric oxide . . . . . . 0.93 per cent Magnesium oxide 0.11 per cent This showed the substance in the pure state to be CSi. The furnace in which these experiments were carried out was made of refractory bricks, the interior dimensions being 10 by 4 by 4 inches. The current was carried by a core of granulated carbon, as shown in Figure 85. 8 C. R. 117, 425, (1893). 8 Journ. of the Franklin Inst. 136, 194 and 279, (1893). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 211 Figure 86 shows an end view of this furnace and the layers of different materials after a run. B is a solid mass of sand FIG. 85. Longitudinal section of Acheson's experimental carborundum furnace and carbon held together by fused salt. C is chief product of the reaction, crystallized carbide of silicon. W represents a white or gray-greenish-looking shell, and consists of small pieces the size of the original grains. They are soft, and may easily be reduced to fine powder, and are of no value as an abrasive, though analysis shows them to be principally carbide of sili- con. It is amorphous carborundum, or carborundum fire sand. G- is , ., , ... , , FIG. 86. Transverse section of graphite mixed with carborundum, Aches0 n's experimental carbo- and D is the core, only , portion of rundum furnace which becomes graphitized even though used repeatedly. The output of this small furnace was \ pound a day. 10 The furnaces used at Monongahelainl893 were 18 inches wide, 12 inches deep, and 6 feet long. The core was of granular carbon in the form of a sheet 10 inches wide, 1 inch deep, and 5J feet long. In 7| to 8 hours a portion of the charge was transformed into 50 pounds of crystallized carborundum. On moving to Niagara Falls the furnaces were construe ted as shown in Figure 87. n The end walls are built of refractory 10 FitzGerald, Journ. Franklin Inst. 143, 81, (1897). 11 FitzGerald, Carborundum, p. 8. . 212 APPLIED ELECTROCHEMISTRY brick and clay, and carry electrodes, 5 2 , consisting of rectangu- lar carbon rods clamped together. Contact is made with the copper cables by the copper plates, > 5 , as shown. A are the brick side walls of the furnace put together without cement. D is the mixture, C the core of granulated carbon, and c is fine carbon powder for the purpose of making contact between the carbon electrodes and the core. Up to 1907 the total length of this furnace was 7 meters ; the inside dimensions were, length, 5 meters, width, 1.8 meters, and height, 1.7 meters. The elec- trodes consisted of 25 carbon rods, 86 centimeters in length, and 10 by 10 centimeters in cross section. The core was 53 FIG. 87. Longitudinal section of carborundum furnace centimeters in diameter. A perspective of the furnace in oper- ation is shown in Figure 88. The power absorbed by each furnace is 746 kilowatts. The voltage varies from 210 volts at the start to 75 volts when the resistance of the core had dropped to its final constant value. Soon after the current is turned on, carbon monoxide is pro- duced, due to the oxidation of the carbon in the core and in the charge. The gas is always lighted, and burns during the run. When the temperature has become sufficiently high, carbo- rundum is formed according to the following reaction : Si0 2 + 3 C = CSi + 2 CO. The heating lasts 36 hours, and produces 3150 kilograms of crystallized carborundum, surrounding the core to a depth of PRODUCTS OF THE RESISTANCE AND ARC FURNACE 213 214 APPLIED ELECTROCHEMISTRY from 25 to 30 centimeters, This corresponds to 8.5 kilowatt hours per kilogram, which is a great improvement over the first furnaces of the Carborundum Company at Monongahela, which were built for 100 kilowatts, and yielded one kilogram of car- borundum for an expenditure of 17.6 kilowatt hours. The present electrical equipment of the Carborundum Company at Niagara Falls has a capacity of 5300 kilowatts. 12 The raw materials used by the Carborundum Company con- sist of ground quartz 99.5 percent silica, coke, such as is used in blast furnaces, sawdust, and sodium chloride. The object of the sawdust is to make the charge porous to facilitate the escape of the carbon monoxide. The coke used for the core is sifted to get rid of the powder; that used for the charge is powdered. The charge is made up in lots of 500 kilograms, and has the following composition : Quartz 261 kilograms Coke 177 kilograms Sawdust 53 kilograms Salt 9 kilograms 500 In 1907 the furnace plant was remodeled, 13 and the furnaces were made 9.15 meters long and 3.67 meters wide. These are presumably outside dimensions. The power absorbed is now 1600 kilowatts with the maximum current 20,000 amperes. The yield of each furnace in one run is 15,000 pounds, or 6800 kilograms, of crystallized carborundum. On coining from the furnace the carborundum is ground, treated with concen- trated sulphuric acid to remove harmful impurities, and is washed with water. It is then sorted into different sizes. Table 25 gives the production of carborundum in this country and its value including the year 1909. 14 12 Electrochem. and Met. Ind. 7, 190, (1909). 13 Min. Ind. 16, 155, (1907); 17, 112, (1908). " Min. Ind. 18, 86, (1909). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 215 TABLE 25 YEAR METRIC TONS VALUE IN DOLLARS 1891 . 0.023 1892 ..'..... 1 1893 .'-...-. 7 1894 24 1895 . 102 189G 540 1897 564 1898 724 1899 791 1900 . . ' 1089 1901 1742 1902 1698 1903 2160 1904 3203 1905 2539 1906 '. . 2824 1907 3418 1908 2226 1909 2983 366,000 154,000 151,000 157,000 168,000 268,000 261,000 333,000 494,000 391,000 435,000 452,000 294,000 389,000 In 1902 the cost of manufacture was 4 cents to 5 cents a pound, and during this year the average selling price was 10 cents a pound. 15 The only producer in this country is the Carborundum Company of Niagara Falls. In Europe it is produced at La Bathie, France, Iserhohn, Germany, and Prague. 16 Carborundum is used principally as an abrasive and as a sub- stitute for ferrosilicon in the manufacture of steel. In 1902 one third the total output was consumed in this industry. 17 The abrasive qualities of carborundum are affected by its great brittleness, on account of which it will not cut diamond unless reduced to a fine powder. 9 It is made into polishing wheels by mixing with a certain amount of kaolin and feldspar as a binder, compressing in a hydraulic press, and burning in a furnace such as is used in the manufacture of porcelain. Carborundum is 15 Min. Ind. 11, 78, (1902). "Min. Ind. 11, 227, (1902). "Mm. Ind. 10, 253, (1901). 216 APPLIED ELECTROCHEMISTRY also used in wireless telegraphy as a detector, and in a different form, known as Silundum, 1 * 1 us a resistance for heating purposes. Silundum is made by exposing rods of carbon to the vapor of silicon, which penetrates the carbon, changing it to silundum arid thereby increasing its electrical resistance to a sufficient extent to make it a good resistor. In the form of bricks car- borundum is used as a refractory material in building furnaces, when the temperature to be withstood is very high. Silicon carbide is colorless when pure, 10 but the commercial product is black, due either to carbon, iron, or to a" thin film of silica 19 on the surface. The following analysis is due to Moissan : PER CKNT Theoretical Silicon 69.70 69.85 70.00 Carbon 30.00 29.80 30.00 The following is an analysis of Acheson's product : u Silicon 64.93 per cent Carbon and oxygen 33.26 per cent Loss in beating 1.36 per cent Aluminum 0.25 per cent Calcium, magnesium, iron .... trace 99.80 per cent When the same material was purified by hydrochloric acid and sodium hydrate, by heating in oxygen, and finally by heating with hydrofluoric acid, its analysis gave the following result : Silicon 69.10 per cent Carbon 30.20 per cent A1 2 O 3 and Fe 2 O 3 0.49 per cent CaO 0.15 per cent 99.94 per cent The density is 3.2. The crystals have been found by Frazier to be rhombohedral. 20 It easily scratches ruby, and, as stated above, when finely powdered, will polish diamond. re Boiling, Electrochem. and Met. Ind. 7, 25, (1909). w Min. Ind. 16, 155, (1907). 20 Journ. Franklin Inst. 136, 289, (1893). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 217 When carborundum is heated to a sufficiently high tempera- ture silicon is vaporized, leaving carbon in the form of graphite. The temperature at which decomposition takes place has been found by Tucker and Lampen 21 to be 2220 and the temperature of formation, 1950C. There is hardly a doubt that both the reactions, SiO 2 + 3 C = SiC + 2 CO and SiC = Si + C are reversible. The temperature of formation therefore depends on the partialpressure of carbon monoxide, and the temperature of decomposition on the partial pressure of silicon vapor, for according to the Phase Rule each of these systems has one degree of freedom. These values, however, probably represent fairly well the temperatures of formation and decomposition in the Acheson furnace. Carborundum is not attacked by sulphur or oxygen at 1000 C., 7 but according to Acheson it is oxidized in an at- mosphere containing considerable oxygen at 1470 C. 22 It is attacked slightly by chlorine at 600. Fused potassium nitrate and chlorate, boiling sulphuric and hydrofluoric acids are all without action. The same is true of a boiling mixture of con- centrated nitric and hydrofluoric acids. On the other hand it is attacked by fused potassium hydrate, forming potassium carbonate and silicate. 3. SlLOXICON 1 There are a number of compounds, besides the carbide of silicon, that contain carbon and silicon in the same proportions as the carbide. In 1881 Schiitzenberger and Colson 2 prepared a compound of the formula SiCO by heating silicon in an at- mosphere of carbon dioxide. The reaction is stated to be 3 Si + 2 CO 2 = SiO 2 + 2 SiCO. 21 Journ. Am. Chem. Soc. 28, 853, (1906). 22 Electrochem. Ind. i, 373, (1903). 1 The name given by Acheson to compounds of carbon, silicon, and oxygen in varying amounts. 2 C. R. 92, 1508, (1881). 218 APPLIED ELECTROCHEMISTRY The same compound was formed at a higher temperature by the direct union of silicon and carbon monoxide. A compound of the formula Si 4 4 N was formed in a similar way. On heat- ing silicon in a stream of hydrogen saturated with benzene at 50 to 60 C. two compounds were obtained, one of the formula C 2 Si, and the other of a variable composition, but fre- quently containing more oxygen than corresponds to the formula CSiO 2 . 3 On heating silicon in a vapor of carbon sul- phide two compounds deposited in the cold part of the com- bustion tube corresponding to the formulae SiSO and SiS. In the boat containing the silicon a greenish powder was obtained which, when purified by boiling in potassium hydrate and treat- ing with hydrofluoric acid, had the composition Si 4 C 4 S. This when heated in a current of oxygen gave Si 4 C 4 O 2 . These bodies all look alike and can be distinguished only by analysis. 4 They are pale green powders, infusible, unattackable by hydro- fluoric acid or strong solutions of caustic alkali. Fused caus- tic alkali decomposes them, giving alkali silicate and carbonate. They resist oxidation at red heat. It will be seen that these compounds also resemble the compound obtained by Schiitzen- berger 4 in 1892, and which analysis showed to be SiC, though the color of the latter compound is described as a clear green. It, therefore, seems that carborundum exists in two forms, one crystalline and the other amorphous, while the amorphous form has all the appearance of other compounds containing silicon and oxygen in the same proportions as carborundum, together with a variable amount of oxygen. From the contradictory statements 6 found in the literature it seems that the layer of material which is formed just outside the carborundum consists of silicon, carbon, and oxygen in varying amounts, and that it goes by the names of amorphous carborundum, carborundum 8 Colson, C. R. 94, 1316, 1526, (1882). 4 Schiitzenberger, C. R. 114, 1089, (1892). 6 In Min. Ind. 15, 93, (1906), it is stated that another product of the carborun- dum furnace is amorphous carborundum or carborundum fire sand, and that siloxicon is a second product obtained when insufficient coke is present, consist- ing of carbon, silicon, and oxygen, while on p. 96 the statement is made that amorphous carborundum contains carbon, silicon, and oxygen. PRODUCTS OF THE RESISTANCE AND ARC FURNACE 219 fire sand, or siloxicon. The latter name is due to Acheson, who took out a patent for its production in 1903. 6 In the manufacture of siloxicon it is important not to have sufficient carbon in the charge to reduce the silica completely, and to keep the temperature constant within certain narrow limits. For this purpose the furnace is built with more than one core, thus making the distribution of temperature more even. The charge, consisting of one third carbon and two thirds silica, is made up of powdered carbon, powdered silica, and sawdust, the silica and carbon contents of the sawdust being taken into account. The density of siloxicon is 2.7. 7 When heated in an atmos- phere containing a large amount of oxygen to about 1470 C., it is oxidized, giving silica and carbon dioxide, 8 while in the absence of oxygen at a higher temperature it is converted into carborundum. Siloxicon is used to make crucibles and for furnace lining, as it is not attacked by melted metals or by slags. 4. SILICON The manufacture of silicon is now carried out by the Carbo- rundum Company according to patents of F. J. Tone. 1 Arc furnaces are used in which two vertical electrodes ex- tend for a considerable depth into the charge of coke and sand. The furnace is built of fire brick lined inside with carbon. Each furnace has a capacity of 910 kilowatts, and the metal is tapped out at intervals of a few hours in ingots weighing from 600 to 800 pounds. It is made in different grades, varying from 90 to 97 per cent pure. Silicon is used principally in the steel in- dustry in place of ferrosilicon. The production of silicon in 1908 was 600 long tons, valued at $72,000. 2 Previous to its Electrochem. and Met. Ind. 1, 287, (1903). 7 FitzGerald, Electrochem. and Met. Ind. 2, 439, (1904). e Acheson, Electrochem. and Met. Ind. 2, 373, (1904). 1 Electrochem. and Met. Ind. 7, 192, (1909). 2 Min. Ind. 17, 13, (1908). 220 APPLIED ELECTROCHEMISTRY manufacture by the Carborundum Company the price of silicon was $4 a pound. Silicon can also be made in small laboratory furnaces. 3 5. GRAPHITE Graphite was known to the ancients, but up to the time of Scheele no distinction was made between it and the closely similar substance molybdenum sulphide, MoSg. 1 Both leave a mark on paper and were called plumbago on account of the belief that they contained lead. In order to define graphite more definitely, Berthelot 2 proposed that only that variety of carbon be given this name which, on oxidation with powerful oxidizing agents at low temperatures, gives graphitic oxide. Graphitic oxide has different properties, depending on the differences in the graphite from which it is made, but all varieties are insoluble and deflagrate on heating. Amorphous carbon, when oxidized with a mixture of potassium chlorate and fuming nitric acid, the oxidizing agent used by Berthelot, is changed to a soluble substance, and diamond is not affected. This is a method of separating the three different kinds of carbon. The artificial production of graphite by dissolving carbon in cast iron and allowing to cool slowly was first observed by Scheele in 1778. 1 It has since been made by Moissan by dis- solving in iron, as well as in a number of other metals, and by heating pure sugar carbon in the electric arc. 3 Diamond also may be changed to graphite by heating in the electric arc. Despretz, 4 in his work on carbon, produced graphite by heating carbon in an electric furnace. These observations do not agree with those of Acheson, who early in his experience in the manu- facture of carborundum noticed that graphite occasionally formed 8 Tucker, Met. and Chem. Eng. 8, 19, (1910). 1 Roscoe and Schorlemmer, Treatise on Chemistry, 3d ed. Vol. 1, p. 730. 2 Ann. de Chim. et de Phys. (4) 19, 393, (1870). 3 Moissan, The Electric Furnace, p. 61. See also FitzGerald, Kunstlicher Graphite, Vol. 15 of the Engelhardt Monographien. * C. R. 28, 755 ; 281, 48 and 709, (1849). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 221 next to the core, 5 and that when coke from bituminous coal was used for the core quite a large amount of it was converted into graphite, whereas when the purer petroleum coke was used very little was so changed. The greater the amount of impurity in the coke, the larger was the amount of graphite produced. These facts led Acheson to the theory that graphite is not produced by simply heating carbon, but that a carbide must first be pro- duced and then decomposed by a higher temperature, volatilizing the metallic element and leaving the carbon in the form of graphite. The effect of the impurities is catalytic, since the amount of graphite formed was always too great to be accounted for by the simple decomposition of the quantity of carbide cor- responding to the impurity present. If only a small amount of impurity is present, it is lost by volatilization before all the carbon can be graph itized. Acheson also found that the production of graphite was greatly increased by adding a considerable quantity of any substance that could form a carbide, such as silica, alumi- num oxide, lime, or iron oxide. 6 At first the charge was made up with enough impurity to change all the carbon to carbide at once. For example, a charge would consist of 50 per cent coke, with sand, salt, and sawdust. Carborundum was then formed and by heating to a higher temperature the carborundum is decomposed, leaving graphite. It was found, however, that so much carbide-forming element was not necessary and that such substances as anthracite coal that had impurities evenly dis- tributed through them could be converted into very pure graph- ite. 7 This is at present one of the principal kinds of carbon used in this industry. Intimate mixture of carbon and the impurity is not necessary, as the carbide-forming element can be vaporized and caused to penetrate the entire charge, thereby converting it to graphite. 8 Petroleum coke is one form of carbon used in this process. Lumps of the coke are imbedded in powder formed from the same material and 5 per cent of iron oxide is sprinkled in. The iron oxide is reduced, iron is formed at the bottom of the furnace, 5 Journ. Franklin lust. 147, 476, (1899). 7 U. S. Pat. 645,285, (1899). 6 U. S. Pat. 568,323, (1893). U. S. Pat. 711,011, (1900). 222 APPLIED ELECTROCHEMISTRY and as the temperature is raised volatilizes and penetrates the whole charge. A very soft quality of graphite is obtained when the carbide-forming material is more than 20 per cent by weight of the charge, but less than the amount necessary to change all the carbon to carbide at once. 9 The furnaces for graphitizing carbon in bulk have a central core similar to the carborundum furnace. 10 FIG. 89. Section of graphite furnace for rectangular electrodes FIG. 90. Section of graphite furnace for circular electrodes In making graphite into electrodes, crucibles, or other finished products, a mixture of 97 per cent carbon and 3 per cent iron 9 U. S. Pat. 836,355, (1906). 10 Richards, Electrochein. Ind. 1, 54, (1902). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 223 oxide 11 is mixed with a binding material consisting of water and a little molasses, and is molded into the desired form. The molded objects are then dried and placed in the furnace, where they are changed to graphite without altering their shape. Fig- ures 89 and 90 show the methods of arranging rectangular and FIG. 91. Electric furnace in which graphite is made artificially by the International Acheson Graphite Company, Niagara Falls circular electrodes respectively. The base of the furnace consists of bricks, covered with a refractory material, h. The end walls, 6, are of brick and hold the carbon electrodes, c. The bottom of the furnace is covered with a layer of granulated coke about 5 centimeters thick, on which the electrodes are placed in piles at right angles to the axis of the furnace, separated from each other by about one fifth the width of the electrodes. This space is then filled .with granulated coke, g, arid the furnace is covered with a mixture of coke and sand, i. Figure 91 is from a photo- graph of the furnace now used for graphitizing carbon in all forms. u U. S. Pat. 617,029, (1898). 224 APPLIED ELECTROCHEMISTRY The following data are given by FitzGerald : 12 Distance between terminals 360 inches Length of space filled by electrodes 302 inches Length of space filled by granular carbon .... 58 inches Length of electrodes under treatment 24 inches Width of electrodes under treatment 5 inches Height of pile of electrodes 17 inches Initial voltage 210 volts Initial amperage 1400 amperes Final voltage 80 volts Final amperage 9000 amperes In 1902 the plant of the International Acheson Graphite Company consisted of ten furnaces and 1000 available horse power. In 1909 the plant was increased to 22 furnaces and 4000 horse power. 13 The yearly production of manufactured graphite is given in Table 26. 14 " TABLE 26 The Production of Graphite YEAR POUNDS VALUE IN DOLLARS 1897 162,000 10,100 1898 186,000 11,600 1899 406,000 32,500 1900 . . 861,000 68,900 1901 2,500,000 119,000 1902 . . . 2,359,000 111,000 1903 2,620,000 179,000 1904 3,248,000 218,000 1905 4,596,000 314,000 1906 4,868,000 313,000 1907 6,924,000 484,000 1908 7,386,000 503,000 1909 . . 6,871,000 467,000 12 Electrochem. and Met. Ind. 3, 417, (1905). 1 3 Electrochem. and Met. Ind. 7, 187, (1909). i* Min. Ind. 18, 384, (1909). The figures in the table are rounded off. PRODUCTS OF THE RESISTANCE AND ARC FURNACE 225 6. CARBON BISULPHIDE Great improvement has been made in the manufacture of carbon bisulphide by using an electric furnace in place of the small clay or iron retorts which have to be heated ex- ternally. In the old process, only a small fraction of the heat applied to the outside of the retort penetrated to the mixture of carbon and sulphur inside, and the process was so disagreeable on account of small leaks and the high tem- peratures of the retort room that some manufacturers gave it up altogether. E. R. Tay- lor, 1 however, has succeeded in overcoming these difficulties entirely by the use of the fur- nace shown in cross section in Figure 92, patented in 1899 2 and in operation at Perm Yan, New York. This furnace is 12.5 meters high and the diam- eter at the base 4.87 meters. 3 At a height of 3.68 meters the diameter is reduced to 2.5 me- ters for a distance of 4.87 me- ters, where it narrows down to the top for the remaining length. The electrodes are at the base and are four in num- ber, arranged 90 degrees apart. FIG. 92. Taylor's electric furnace for Opposite electrodes are Con- making carbon bisulphide 1 E. R. Taylor, Trans. Am. Electrochem. Soc. 1, 115, (1902) and 2, 185, (1902) 2 U. S. Pat. 688,364, filed 1899, renewed 1901. 3 Haber, f. Elektroch. 9, 399, (1903). Q 226 APPLIED ELECTROCHEMISTRY nected to the same terminal of the alternating current machine. Wear on the electrodes is reduced to practically nothing by covering them with conducting carbon, which acts as the re- sistor. Charcoal is fed in at the top and sulphur through the annular spaces in the walls, thus preventing loss of heat. The sulphur is melted by the heat which would otherwise be lost through the walls, and flows down on to the electrodes, where it is heated to a temperature at which it combines with carbon. The carbon bisulphide vaporizes, passes off through the top of the furnace, and is condensed in cooling coils. The furnace is so tight that no odor is noticeable, and its operation is contin- uous. The production in 1903 was 3175 kilograms per day, with a consumption of 220 horse power 3 and the furnace had been in operation for two and a half years with only one interruption for the purpose of cleaning out. 7. PHOSPHORUS Phosphorus is another product the manufacture of which has been improved by the use of heat derived from electricity. The older method consists in treating calcium phosphate with sulphuric acid, which changes the triphosphate to monophos- phate : Ca 3 (P0 4 ) 2 + 2 H 2 S0 4 = 2 CaSO 4 + CaH 4 (PO 4 ) 2 . The monophosphate is then mixed with carbon and dried, by which it is changed to metaphosphate : CaH 4 (P0 4 ) 2 = Ca(P0 8 ) 2 + 2 H 2 O. The metaphosphate is then heated in small retorts in which the following reaction takes place : 3 Ca(PO 3 ) 2 + 10 C = Ca 3 (PO 4 ) 2 + 10 CO + 4 P. This process is imperfect in that a portion of the phosphorus is changed in the last operation to the product with which the operation is begun. Wohler proposed the use of silica and carbon, by which all the phosphorus would be recovered, as shown by the following reaction : Ca 3 (PO 4 ) 2 + 3 SiO 2 + 5 C = 3 CaSiO 3 + 5 CO + 2 P, PRODUCTS OF THE RESISTANCE AND ARC FURNACE 227 but it was never successful till the introduction of the electric furnace, on account of the difficulty of obtaining the necessary temperature and of finding vessels to withstand it. 1 In 1889 the use of electric furnaces for the manufacture of phosphorus was patented by J. B. Readman. 2 The process does not seem to have been immedi- ately employed on a large scale, however. In 189T the firm of Allbright and Wilson built works at Niagara Falls, using 300 horse power, for making phosphorus in the fnr The Roadman-Parker electric furnace for produc- ing phosphorus The furnaces FlG . 93. Vertical FIG. 94. Horizontal are illustrated in Fig- section section ures 93 and 94. Each produces 170 pounds a day. Over half the world's production of phosphorus is now made in electric furnaces. 4 8. ALUNDUM Fused aluminum oxide, chemically identical with corundum, has received the trade name of Alundum. The process for making this abrasive in the electric furnace was patented in 1900 by C. B. Jacobs. 6 His furnace was rectangular in shape, made of sheet iron and brick, and was lined inside with car- bon. An arc was formed between four pairs of electrodes near the movable bottom of the furnace. As the aluminum oxide fused and covered the bottom of the furnace, it was gradually lowered, thereby making a layer of fused aluminum oxide which cooled slowly. This process gives the abrasive a hard- ness greater than corundum. i Min. Ind. 14, 494, (1905). Min. Ind. 6, 637, (1897), 7, 557, (1898). 2U. S. Pat. 147,943, (1889). *Min. Ind. 9, 768, (1900). 6 U. S. Pat. 659,926, (1900). 228 APPLIED ELECTROCHEMISTRY The Norton Emery Wheel Company of Worcester are the sole manufacturers of alundum. Their factory is at Niagara Falls. Bauxite, the raw material, is dehydrated before feeding into the furnaces. The yearly production is given in Table 27. 6 TABLE 27 Production of Alundum YEAR POUNDS VALUE IN DOLLARS 1904 4 020 000 281 400 1905 3 612 000 259 840 1906 4,331,000 303 190 1907 6,751,000 405,090 1908 ... 3,160,000 189,600 1909 13,758,000 814,680 9. ALUMINUM With the exception of silicon and oxygen, aluminum is the most widely distributed element in nature, 1 occurring princi- pally as silicates in clays. Only a limited number of its com- pounds can be used for extracting aluminum, however, chief among which is bauxite, A1O 3 H 3 . The name aluminum is derived from alumen, a term applied by the Romans to all bodies of astringent taste. The attempts to isolate aluminum date from 1807, when Davy was unsuccessful in applying to this problem the method em- ployed in isolating the alkali metals. Oersted seems to have made aluminum in 1824 by heating the chloride with potassium amalgam. Wohler in 1827 obtained aluminum by decompos- ing the anhydrous chloride with potassium, and in 1864 Bunsen and Deville obtained it independently by the electrolysis of fused aluminum chloride. Previous to the production by the method of electrolysis now used, the halide salts were the source of the metal and were reduced by metallic sodium. Alumina can be reduced by carbon to metallic aluminum by 6Min. Ind. 18,25, (1909). 1 Thorpe, Die. of Chem. 1, 63, (1890). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 229 heating to a temperature above 2100 C., 2 but it is always mixed with aluminum carbide, from which it can be removed by remelting, and obtained in a compact form. This is evidently not a method of making aluminum that could be satisfactorily carried out commercially. If, however, a metal such as copper is added to the mixture, the aluminum can be obtained as an alloy with this other metal. This process was patented in 1884 by the Cowles brothers. 3 The cheap production of pure aluminum, however, was made possible by the discovery of C. M. Hall 4 that alumina, dissolved in a molten mixture of aluminum fluoride and the fluoride of another metal, forms an electrolyte which may be decomposed by an electric current, liberating aluminum at the cathode and oxygen at the anode. Hall's original patent specifies a mixture of 169 parts by weight of aluminum fluoride and 116 parts of potassium fluoride, corresponding to the formula K 2 A1 2 F 8 , and states that this may be made more fusible by replacing part of the potassium fluoride by lithium fluoride, or by simply adding the latter to the above mixture. Another receipt is 84 parts of sodium fluoride to 169 of aluminum fluoride, which may be made by adding aluminum fluoride to cryolite, a mineral of the com- position A1F 8 3 NaF. He placed the carbon-lined crucible in a furnace, melted the mixture, added alumina, and electrolyzed with an anode of copper or carbon. Copper is said to be covered with an oxide which protects it from further action. Subsequent patents show that these mixtures worked well at first, but became less efficient after being electrolyzed some time. A dark substance formed which interfered with the electrolytic action, increased the resistance, and necessitated a change of the bath. This Hall attempted to overcome by using a bath of calcium and aluminum fluoride of the com- position 2 A1F 3 ' CaF 2 . 6 This increases the density to such an extent that the aluminum floats to the surface. It evidently 2 Hutton and Petavel, Phil. Trans. 207, 421, (1907) ; Askenasy and Lebedeff, Z. f. Elektroch. 16, 565, (1910). 8 U. S. Pat. 319,795, (1884). Also Proc. Soc. of Arts, 1885-1886, p. 74. * U. S. Pat. 400,664 and 400,766, filed 1886. U. S. Pat. 400,664, filed 1888. X 230 APPLIED ELECTROCHEMISTRY was not satisfactory, for subsequently a bath made up of 234 parts calcium fluoride, 421 parts cryolite, 845 parts aluminum fluoride, and 3 to 4 per cent calcium chloride was patented. 6 It was claimed that the chloride prevented the clogging of the bath even when in continuous operation. It is evident the dark color must have come from carbon, as no clogging occurred with any of the baths when a metal was used as cathode. 7 In this case, of course, an alloy of aluminum would be obtained. As carried out on a large scale, the crucibles were never heated externally, but simply by the passage of the current it- self. This double use of the current to keep the bath melted and to electrolyze at the same time was patented by Charles S. Bradley. 8 In describing his process, cryolite is considered the electrolyte. The two patents of Hall and Bradley taken to- gether represent the process as actually carried out. In 1887 Paul Heroult patented a very similar process for producing aluminum alloys. 9 This process consisted in fusing pure alumina and keeping it in the fused state by the current, which at the same time decomposes the oxide electrolytically. The cathode is a melted metal, with which the aluminum is to be alloyed, and the anode is carbon. Serious objections were found to using any flux. Among those tried and discarded was cryolite. The patent states that satisfactory results were ob- tained with a carbon crucible 20 centimeters in depth and 14 centimeters in diameter at the top, a carbon anode 5 centimeters in diameter, and a current of 400 amperes at from 20 to 25 volts. This voltage is four or five times that specified by Hall. Brad- ley's patent for the simultaneous use of the current for electrol- ysis and heating was therefore earlier than Heroult's, and as it is stated in Heroult's patent that he had failed to get good re- sults when any flux was mixed with the aluminum oxide, there is no question of priority over Hall's patents. It does not seem, therefore, that the statement often met with, that the processes s U. S. Pat. 400,666, filed 1888. 7 U. S. Pat. 400,667, filed 1888. 8 U. S. Pat. 464,933, filed 1883, granted 1891. U. S. Pat. 387,876, filed December, 1887. PRODUCTS OF THE RESISTANCE AND ARC FURNACE 231 of Hall and Heroult are identical, is borne out by the patents. 10 The Hall patents for the composition of the bath expired in 1905 and the Bradley patents in 1909. u The only producer of aluminum in this country is the Alumi- num Company of America, previous to 1907 known as the Pitts- burg Reduction Company. 12 This company controls three plants, situated at Niagara Falls, Massena, New York, and Shawinegan Falls, Canada. These plants were enlarged in capacity in 1907 to 40,000 horse power, 20,000 horse power, and 15,000 horse power respectively. 13 The six European companies producing aluminum show a maximum consumption of 97,500 horse power. 13 The furnaces used by the American company consist of cast iron troughs lined with carbon. 14 The anode is composed of 48 carbon rods 3 inches in diameter and 15 inches long, manufac- tured by the aluminum company for its own use. 12 Each fur- nace takes about 10,000 amperes at about 5.5 volts. The yield is 1.75 pounds of aluminum per horse power day. 1 * The metal sinks to the bottom and is drawn off, while alumina is thrown in as it is used up. The temperature of the bath may be in- ferred from the following melting points of mixtures of cryolite and alumina. 16 TABLE 28 Melting Points of Mixtures of Cryolite and Alumina PEE CENT CRYOLITE M. P., DEGREES CENTIGRADE PER CENT CRYOLITE M. P., DEGREES CENTIGRADE 100 1000 92 992 97 974 90 980 96 960 85 994 95 915 80 1015 94 960 93 982 10 See for example Pring, Some Electrochemical Centres, p. 26 (1908). " Min. Ind. 17, 23, (1908). 18 Min. Ind. 6, 11, 15, (1907). 12 Min. Ind. 15, 11, (1906). u Min. Ind. 14, 15, (1905). is Pyne, Trans. Am. Electrochem. Soc. 10, 163, (1906). 232 APPLIED ELECTROCHEMISTRY The production of aluminum in the United States and Canada is given in Table 29. 16 TABLE 29 Production of Aluminum in the United States and Canada YEAR POUNDS VALUE IN DOLLARS VALUE PER POUND IN DOLLARS 1897 4,000,000 1,400,000 0.35 1898 5,200,000 1,690,000 0.33 1899 6,500,000 2,113,000 0.33 1900 7,150,000 2,289,000 0.32 1901 ....... 7,150,000 2,238,000 0.31 1902 7,300,000 2,285,000 0.31 1903 . . . 7,500,000 2,325,000 0.31 1904 7,700,000 2,233,000 0.29 1905 11,350,000 3,632,000 0.32 1906 14,350,000 5,166,000 0.36 1907 ..... 26,000,000 10,920,000 0.42 1908 13 000 000 4,095,000 0.32 1909 15,000,000 3,345,000 0.22 The total production of the world for 1909 is estimated at 24,200 metric tons, or 53,300,000 pounds. The cost of manu- facture excluding amortization is said to be about 15 cents a pound. 16 On reading a description of the different expedients patented by Hall to prevent the baths from clogging, becoming discolored, and ceasing to operate properly, it is not surprising that diffi- culties are encountered on attempting to use the reduction of aluminum as a laboratory experiment. Haber and Geipert 17 succeeded in a few runs, though in the last run they met with irregularities. The immediate difficulty that stops an experi- ment on a small scale is a polarization at the anode, due to a thin film of gas, 18 which reduces the current to such a point that the bath freezes up. If a higher voltage is applied it heats the is Min. Ind. 18, 17, (1909). IT Z. f. Elektroch., 8, 1, and 26, (1902). i 8 Thompson, Electrochem. and Met. Ind. 7, 19, (1909). Also Neumann and Olsen, Met. and Chem. Eng. 8, 185, (1910). PKODUCTS OF THE RESISTANCE AND ARC FURNACE 233 bath too much locally and burns up the aluminum. By the use of an anode with a large area this can be prevented to a certain extent. 18 One of the principal uses for aluminum is in the iron and steel industry as a reducing agent. 19 As is well known, it has replaced copper, tin, and brass to a great extent in the manu- facture of a large number of objects in which lightness is desired. 10. SODIUM AND POTASSIUM Sodium and potassium were first isolated by Davy l by electro- lyzing the corresponding fused hydrates. In this process sodium is liberated at the cathode while the negatively charged hydroxyl ion is liberated at the anode. Two of these ions when discharged react together according to the reaction: 2 OH = H 2 O + O. A certain amount of metallic sodium dissolves in the hy irate, diffuses to the anode, and coming in contact with the water reacts to form hydrate with the liberation of hydrogen. 2 It is therefore possible to have both hydrogen and oxygen evolved at the anode, resulting in explosions. At the same time sodium peroxide (Na 2 O 2 ) is formed. The water formed at the anode is not driven off by the temperature of the bath ; on the contrary it has been found that very moist air is dried to a certain extent in passing through the melted hydrate. 2 The apparatus nearly universally used for the production of sodium and potassium is due to Hamilton Young Castner 3 and is shown in Figure 95. It consists in a cast-iron box with an iron cathode, H, insulated from the box and held in an iron pipe fastened into the bottom of the cell. The space between the pipe and electrode is filled with melted hydrate which is allowed to solidify before the electrolysis is begun. Surrounding the 19 For a detailed account of the various purposes to which aluminum is applied, see A. E. Hunt, Journ. Franklin Inst., Vol. 144, (1897). 1 Phil. Trans., 1808, pp. 5 and 21. 2 Lorenz, Elektrolyse Geschmolzener Salze, I, 25, (1905). 8 U. S. Pat. 452,030, filed 1890. 234 APPLIED ELECTROCHEMISTRY cathode is a fine iron gauze diaphragm, M, outside of which is the iron anode, F. The metal is liberated on the cathode and floats to the surface of the hydrate, where it collects in an iron cylinder forming a continuation of the diaphragm. It is re- moved by an iron spoon with fine perforations, which allow the hydrate to drain off, but which holds the metal. The hydrate is added as it is used up, and the process is continuous. An important point is to maintain the temperature as low as pos- sible, not over 20 above the melting point of the hydrate. The FIG. 95. Castner's cell for producing sodium and potassium higher the temperature the less the yield in metal, due of course to its greater solubility in the melted hydrate. As the temper- ature increases, the yield becomes less, until it finally reaches zero. At best the current efficiency is said to be only about 45 per cent. 4 In the patent gas heating is provided, though it is stated that the current can be so regulated as to keep the proper temperature without external heating. There are other processes very similar to that of Castner, some of which are in use, 6 which will be omitted as presenting 4 Ashcroft, Trans. Am. Electrochem. Soc. 9, 123, (1906). 6 See H. Becker, Die Elektrometallurgie der Alkalimetalle, p. 52, (1903). PRODUCTS OF THE RESISTANCE AND ARC FURNACE 235 no new principles; but the principle of the following process, due to Ashcroft, 4 will be described because of its novelty and in spite of the fact that it does not seem as yet to have been carried out on a commercial scale. Melted sodium chloride is electrolyzed with a lead cathode. The lead sodium alloy formed is let into another cell containing melted sodium hydrate. Here the lead alloy acts as the anode and forms sodium hydrate with the hydroxyl ions liberated on its surface, thus avoiding the formation of water and oxygen. At the cathode sodium is lib- erated and removed. To decompose the chloride 7 volts are required, and 2 volts for the hydrate when this anode is used. The voltage is therefore about twice that required in the Castner cell; but as the current efficiency is about 90 per cent, or twice that in the Castner process, the yield per unit of power is the same in the two cases. The advantages claimed by Ashcroft are shown in the following table: ASHCKOFT PROCESS CASTNER PROCESS 0.5 cent 5 cents Cost of power per pound of sodium . . per pound 1 to 5 cents 1 cent per pound 1 to 5 cents 2 cents Upkeep and standing charges .... 2.5 cents 2 cents Total 5 to 9 cents 10 to 14 cents per pound per pound The saving comes in the greater cheapness of the raw material, and there would be a further saving in the value of the chlorine produced. The world's production of sodium in 1907 is estimated at from 3500 to 5000 tons. 6 In the United States there are two com- panies producing about 2000 tons a year. The Electrochem- ical Company at Niagara Falls uses the Castner process, while the Virginia Electrolytic Company at Holcomb Rock, Virginia, is said to employ a process in which fused sodium chloride is electrolyzed. e Min. Ind. 17, 772, (1908). 236 APPLIED ELECTROCHEMISTRY A large part of the sodium made is consumed in the manu- facture of sodium cyanide and sodium peroxide. The process for cyanide 7 consists in passing ammonia over the metal heated in an iron retort to 300 to 400 C., forming sodamide : 2 Na + 2 NH 3 = 2 NaNH 2 + H 2 . This is then treated with charcoal previously heated to redness, giving the cyanide NaNH 2 + C = NaCN + H 2 . A recent purpose to which the metal has been put is the dry- ing of transformer oils. Ashcrof t 4 believes a reduction in the price may increase its uses materially, such as making primary cells, obtaining hydrogen by the decomposition of water, and even for transmitting electric power. The specific conductiv- ity is only about one third that of copper, 8 but if equal weights of metal are considered between two given points, the conduc- tivity would be three times that of copper, as the density of copper is about nine times that of sodium. Some experiments have actually been carried out in power transmission with the sodium protected in iron pipes. 9 11. CALCIUM Calcium was first isolated by Davy in 1808, by combining the methods previously used by him with those of Berzelius and Pontin. 1 Lime was mixed with red oxide of mercury, slightly moistened and placed on a piece of platinum. A glob- ule of mercury in a cavity at the top acted as negative elec- trode, giving on electrolysis an amalgam of calcium, from which the mercury was distilled. Bunsen 2 obtained calcium in very small quantities contain- ing a little mercury by electrolyzing with a high current density a boiling concentrated solution of calcium chloride 7 Roscoe and Schorlemmer, 2, 276, (1907). 8 Landolt-Bornstein Tables, 3d ed. Belts, Min. Ind. 15, 688, (1906) and El. World, 48, 914, (1906). 1 Alembic Club Reprints, No. 6, p. 48, Ostwald Klassiker, No. 45. 2 Fogg. Ann. 91, 623, (1854), in an article on the preparation of chromium. PRODUCTS OF THE RESISTANCE AND ARC FURNACE 237 acidified with hydrochloric acid. The cathode was amalga- mated platinum wire. Rathenau 3 was first to obtain calcium in a compact form in fairly large quantities by a rather original method. The bath consists of calcium chloride very little above its melting point. An iron rod is used as cath- ode, which just touches the surface of the bath. As the melting point of calcium is a little higher than that of C the bath, it solidi- fies on depositing and adheres to the rod, which is grad- ually raised, thus drawing out a stick of calcium with a certain amount of chloride adhering to it. The melt- ing point of the electrolyte may be lowered by adding FIG. 96. Cell of Seward and von Kiigelgen for the pro- calcium fluoride. duction of calcium The anode may be a carbon crucible in which the salt is contained, 4 though Rathenau does not specify his arrangement. The ex- perience of the author has been that this is a much better plan than that adopted by P. Wohler, 5 where the salt is held in an iron vessel and a carbon anode dips into the bath. Due to the 8 Z. f. Elektroch. 10, 608, (1904). * J. H. Goodwin, Proc. Am. Phil. Soc. 43, 381, (1904). 6 Z. f. Elektroch. 11, 612, (1905). 238 APPLIED ELECTROCHEMISTRY high anode current density in this case, the gas is more likely to stop the current by polarization. The heat due to the cur- rent is sufficient to keep the salt melted. Calcium is made in this country only by the Virginia Elec- trolytic Company at Holcomb Rock, Virginia. 6 The process is supposed to consist 7 in electrolyziiig melted calcium chloride in a cell patented by Seward and von Kiigelgen, 8 shown in Figure 96. This cell consists of a circular iron box, J., through the bottom of which projects a conical iron cathode, B, insulated from the box by insulating material, aa. The anode, (7, is a car- bon lining also insulated from the iron box. Above the cathode and concentric with it is a water-cooled collecting ring, E, which separates the metal rising to the surface from the chlorine. The metal accumulates till the ring is full. The top layer is solid, due to the cooling of the air, and the bottom is soft or melted. The solid part is fastened to a hook, F, and gradually drawn out. The production of calcium by the Virginia Electrolytic Com- pany in 1907 was 350 pounds, valued at $613, and about the same amount was produced in 1908. 6 6 Min. Ind. 17, 99, (1908). 'Min. Ind. 16, 131, (1907). U. S. Pat. 880,760. CHAPTER XIII THE ELECTROMETALLURGY OF IRON AND STEEL 1. GENERAL DISCUSSION BEFORE giving an account of the application of electric heat- ing to the iron and steel industry, a short sketch of the older methods of winning and refining iron will not be out of place. The extraction of iron from its ores, consisting principally of oxides of iron mixed with clay, silica, and other impurities, is accomplished by reducing the ore with some form of carbon, usu- ally coke. This operation is carried out in a blast furnace, a cir- cular brick structure lined with silicious brick, and varying in size from 48 feet to 106 feet in height, and from 8 feet to 15 feet in diameter at the base. Figure 97 shows the elevation of a blast furnace. It consists of three principal parts : (1) the cru- cible or hearth at the base, cylindrical in shape, (2) the bosh directly above, which gradually widens, and (3) the stack, from which point the furnace contracts for the rest of its height. The furnace is filled with alternate layers of ore, coke, and flux, the latter usually consisting of calcium carbonate. The object of the flux is to form a fusible slag with the constituents of the ore which are not reduced by the carbon, such as silica and alumina. The heat necessary to raise the charge to a temper- ature high enough for reduction is produced by the combustion of the coke in the charge, by means of air forced in through the tuyeres, F, projecting through the wall of the furnace just below the bosh. The carbon therefore serves the double purpose of furnishing the heat and of reducing the ore. The highest temperature of the furnace is near the tuyeres and a few feet above them ; in this region the slag and iron melt and drop into the crucible, where they separate, the slag 239 240 APPLIED ELECTROCHEMISTRY floating on the iron. These are drawn off from time to time through the tap holes Or and H, and fresh material is fed into the top of the furnace by mechanical means. The iron thus FIG. 97. Elevation of blast furnace produced is known as pig iron, and contains from three to four per cent of carbon, as much as four per cent of silicon, and one per cent of manganese, and a few hundredths of one per cent of sulphur and phosphorus. Only about 23 per cent of the pig iron made in this country is used without subsequent purifica- tion. 1 Purification or refining of iron is accomplished by oxi- dizing the impurities and causing them to form a slag, which floats on the iron. 1 Stoughton, The Metallurgy of Iron and Steel, p. 52. THE ELECTROMETALLURGY OF IRON AND STEEL 241 One method of refining consists in blowing air through the liquid metal in a Bessemer converter. The lining of the con- verter may be either basic, consisting of calcined dolomite (cal- cium and magnesium oxides), or acid, consisting of silica. The Bessemer method is very rapid, silicon and manganese oxidizing in about four minutes from the time when the air is first blown in. The carbon then begins to oxidize to carbon monoxide, which boils up through the metal and comes out of the con- verter in a long flame. In about six minutes from the time the carbon begins to oxidize, it is reduced to approximately 0.04 per cent, and the operation is then stopped. The temperature is higher at the end of the process than at the start, due to the heat of oxidation of the impurities. A calculated amount of carbon is then added, also 1.5 per cent of manganese to remove the oxygen, and 0.2 per cent of silicon to remove the other gases. The steel is then cast into molds. The second method of refining is known as the open hearth or Siemens-Martin process. This consists in melting the pig iron in a large reverberatory furnace, whose lining may be either basic or acid. The oxidation of the impurities is brought about by the excess of oxygen in the furnace gases over that neces- sary to burn the gases. A much longer time is required for purification by the open hearth than by the Bessemer process. In the basic open hearth process enough lime is added to form a very basic slag, which, unlike an acid slag, will dissolve phos- phorus. The lining must also be basic to prevent its being eaten away by the basic slag. The third method of purification is known as the puddling process, in which the iron is melted on the hearth of a rever- beratory furnace lined with oxides of iron. The pig iron is charged by hand through the doors of the furnace and is melted as quickly as possible. During melting, silicon and manganese go into the slag, as well as some of the oxide of the lining. Iron oxide is then added in order to make a very basic slag ; the charge is thoroughly mixed, and the temperature is lowered to the point where the slag begins to oxidize the phosphorus and sulphur before the carbon. After the removal of these 242 APPLIED ELECTROCHEMISTRY impurities, the carbon begins to oxidize and comes off as carbon monoxide, which burns on coming in contact with the air. During this time the puddler stirs the charge vigorously with a long iron rabble, an instrument shaped like a hoe. As the iron becomes pure, its melting point rises and it begins to solidify, since the temperature of the furnace is below the melting point of pure iron. The iron is finally removed in the form of a ball dripping with slag, and is put through a squeezer to remove the slag as much as possible. This product is known as wrought iron. It is converted into steel by two methods, (1) the ce- mentation, and (2) the crucible process. In the cementation process the wrought iron is carburized by heating, without melting, in contact with carbon. The carbon slowly penetrates the iron and changes it to steel. In the crucible process the wrought iron is cut up into small pieces and is melted in covered crucibles with the desired amount of carbon or other element that is to be alloyed with it. When the process is finished the steel is cast into molds. By thus remelting the iron, the slag is removed and the required amounts of carbon, silicon, and manganese are added. 2. THE ELECTROTHERMIC REDUCTION OF IRON ORES The conditions under which electric heating can economically be substituted for the heat of combustion of coke in the reduc- tion of iron ores are purely local. In places where iron ore can be obtained cheaply, where metallurgical coke is expensive, where water power is cheap, and where iron would have to be hauled from a great distance to supply the local demand, it may be possible to produce iron by electric heating at a price low enough to compete with that brought from a distance. These conditions exist in Canada, Sweden, and California. 1 The first attempt to apply electric heating to the metallurgy of iron was made in 1853 by Pinchon, 2 and in 1862 Monkton took a patent in England for the reduction of ores by the 1 Eugene Haanel, Trans. Am. Electrochem. Soc. 15, 25, (1909) l and P. McN. Bennie, ibid. p. 35. 2 B. Neumann, Electrometallurgie des Eisens, p. 3, (1907). THE ELECTROMETALLURGY OF IRON AND STEEL 243 FIG. 98. Stassano's first furnace at Rome electric current. Sir Wil- liam Siemens again called attention to this subject in a lecture before the Society of Telegraph En- gineers in London in 1880. 3 The first, however, to show by experiments on a large scale that iron can be re- duced commercially by electric heating was the Italian army officer, Major Stassano. 4 Patents were taken out by him in the year 1898 in different countries, con- sequently this date marks the beginning of the actual appli- cation of electricity to the metallurgy of iron. The contrac- tion of the carbide industry in 1899 to 1900, due to over- production, leaving idle a number of water-power stations in southeastern France, for which some new application of electric power was needed, also hastened the introduction of electric heating in the iron industry. 6 Stassano's prelim- FIG. 99. Horizontal section of Stassano's electric furnace at Darfo Elektrotech. Z. 1, 325, (1880). 4 Askenasy, Technische Elektrochemie, 94, (1910). 6 J. B. C. Kershaw, Electrometallurgy, p. 175, (1908). 244 APPLIED ELECTROCHEMISTRY inary experiments on the reduction of iron ore were carried out at Rome in 1898, 6 with the 150 horse power furnace repre- sented in Figure 98. It is seen to resemble an ordinary blast furnace. Since there was no combustion of carbon, no reduc- ing gases were produced ; consequently, in order to bring the FIG. 100. Vertical section of Stassano's electric furnace at Darfo carbon and ore in intimate contact, they were powdered, mixed, and made into briquettes with pitch as a binder. The furnace 6 See an article by Stassano reprinted in Haanel's Keport, p. 178, (1904). THE ELECTROMETALLURGY OF IRON AND STEEL 245 was first heated without a charge ; an iron grating was then placed in the furnace 20 centimeters above the arc, and the mix- ture was charged in from the hopper at the top and was held up by the grating. The grating eventually melted, and the ore in contact with it was reduced. In this state the mixture which lay on the grating became fused and formed an arch, which supported the charge even when the grating melted away. As the heat from the arc penetrated the mass above the arch, iron was reduced and dropped into the crucible below. In the course of twelve hours the arch increased so in thickness, due to the slag produced, that it prevented the efficient heat- ing of the charge above. Consequently this form of furnace was given up, and one was adopted in which the material was- introduced below the arc, as is done in refining furnaces. The final form adopted at Darfo, in northern Italy, is shown in Figures 99 and 100. Movement of the entire chamber in which the fu- sion takes place is effected by rotating about an axis inclined to the vertical. The electricity is conducted to the furnace by sliding contacts on two metal rings at the top of the furnace. This, furnace worked per- fectly satisfactorily, even when run for several days. The most difficult ques- tions to decide were the re- lation between the size of the cavity and the energy to be supplied, and the manner of making the refractory lining. The carbon electrodes were 1.5 meters long and lasted sixty consec- utive hours. The furnace was supplied with 1000 amperes at 100 volts, and since the value of the cosine of the phase differ- ence between electromotive force and current was 0.8, the power FIG. 101. The Keller electric furnace for reducing iron ore 246 APPLIED ELECTROCHEMISTRY consumed was 80 kilowatts. The best yield with this furnace was one kilogram of soft iron for 3.2 kilowatt hours, and the iron obtained was always over 99 per cent pure. The ore, which was from the island of Elba, had the following com- position : Fe 2 O 3 93.020 per cent MnO 0.619 per cent SiO 2 3.792 per cent CaO, MgO 0.500 per cent Sulphur , 0.058 per cent Phosphorus 0.056 per cent Moisture 1.720 per cent According to Stassano, the plant at Darfo was shut down for reasons not directly connected with the success of the process. The Keller furnace for making pig iron is shown in Figure 101. This furnace was seen in operation by the Canadian Commission at Livet, France, in 1904. It T consists of two iron castings of square cross section, forming two shafts com- municating with each other at their lower ends by a lateral canal. The castings are lined with refractory material. The base of each shaft is provided with a carbon block, these two blocks being connected to each other outside the furnace by copper bars. On starting, before there is metal in the canal, the current flows from one block to the other through the copper bar, but when enough metal has been reduced to partially fill the canal, most of the current flows through the melted metal. The electrodes are 1.4 meters long and 85 by 85 centimeters in cross section. The cost of electrodes per metric ton of pig iron is estimated by Keller at 3.85 francs. The energy absorbed per metric ton of pig iron in a furnace supplied with 11,000 amperes at 60 volts was 0.390 kilowatt year for the run, and with a smaller furnace supplied with 7000 amperes at 55 volts it was 0.186 kilowatt year for the run. 8 Haanel's Report, p. 15, (1904). Haanel's Report, p. 20, (1904). THE ELECTROMETALLURGY OF IRON AND STEEL 247 .EXPERIMENTAL. ElZCTRICfURNACt Following the tour of inspection by the Canadian Commission, an investigation was carried out for the Canadian government in 1906 by Heroult, to see (1) whether magnetite could be eco- nomically smelted by the electrothermic process ; (2) whether ores containing sulphur, but not manganese, could be made into pig iron of marketable composition ; and (3) whether charcoal could be substituted for coke. The furnaces were slightly modified as the investigation proceeded, and the final form is shown in Figure 102. It consists of a cylindrical iron casting | inch thick, bolted to a bottom plate of cast iron 48 inches in diameter. The casting was made in two sections bolted together by angle irons. In order to make inductance small, the magnetic circuit was broken by re- placing a vertical strip of 10 inches width in the casting by copper. Rods of iron were cast into the bottom plate to secure good contact with the car- bon paste rammed into the lower part of the furnace. The electrodes, 6 feet long and 16 by 16 inches in cross section, were manufactured by a pro- cess of Heroult's and were imported from Sweden. The pipe k was for the purpose of cooling the electrode holder by a current of air. The current was between 4000 and 5000 amperes at 36 to 39 volts, and the power factor was 0.919. The ores used in the experiments below were of the following composition: FIG. 102. Heroult experimen- tal furnace at Sault Ste. Marie, for reducing iron ore 248 APPLIED ELECTROCHEMISTRY TABLE 30 Composition of Ores investigated by Heroult for the Canadian Government HEMATITE MAGNETITE ROASTED PYRRHOTITE TlTANIPEROUS IRON ORE 1 2 3 Fe 62.23 56.69 55.85 58.29 45.80 43.59 SiO 2 5.42 6.20 6.60 4.00 10.96 7.12 Fe 2 3 FeO A1 2 3 88.90 55.42 23.04 2.56 60.74 17.18 1.48 55.31 25.20 2.24 65.43 3.31 30.30 28.78 7.00 2.51 CaO 0.61 2.00 2.48 2.40 3.92 1.00 MgO Mn 0.30 0.16 6.84 5.50 4.00 3.53 4.14 MnO 0.258 0.13 P 0.044 0.01 0.016 0.415 0.016 0.028 S 0.0002 0.05 0.57 0.45 1.56 0.04 CO 2 and unde- termined Loss on ignition Cu Ni 3.609 4.923 M6 2.23 2.48 TiO 2 17.82 100.426 100.00 100.00 100.00 99.648 The consumption of the electrode in these experiments was 8.9 kilograms per metric ton of pig iron produced. The yield per unit of energy vafried somewhat, but was approximately 0.25 kilowatt year of 365 days per metric ton of pig iron. The results of these experiments were : 1. Canadian ores, chiefly magnetites, can be as economically smelted as hematites by the electrothermic process. 2. Ores of high sulphur content can be made into pig iron containing only a few thousandths of one per cent of sulphur. 3. The silicon content can be varied as required for the class of pig iron to be produced. THE ELECTROMETALLURGY OF IRON AND STEEL 249 4. Charcoal which can be cheaply produced from mill refuse or wood which could not otherwise be utilized, and peat coke, can be substituted for coke without being briquetted with the ore. 5. A ferro-nickel pig can be produced practically free from sulphur, and of fine quality, from roasted nickeliferous pyrrhotite. 6. Titaniferous iron ores containing up to five per cent can be successfully treated by the electrothermic process. These results demonstrated the feasibility of applying the electrothermic process to the reduction of iron ores. 9 All that was necessary to put it on a commercial basis was the construction of a furnace that could be economically and suc- cessfully used in prac- tice. This was under- taken by three Swedish engineers, Messrs. Gron- wall, Lindblad, and Stal- hane, at Domnarfvet, Sweden. They concen- trated their attention on the construction of a furnace following the suggestions contained in the report of Heroult's experiments for the Canadian government, which were, (1) charg- ing by labor-saving ma- chinery, (2) collection and use of carbon monoxide produced by the reduction of the ore, (3) automatic regulation of electrodes, and (4) a sufficiently fire BricA 3 Masncsitt FIG. 103. Electric furnace at Domnarfvet, Swe- den, for reducing iron ore 9 Haanel, Trans. Ain. Electrochem. Soc. 15, 25, (1909). 250 APPLIED ELECTROCHEMISTRY high shaft containing the charge to permit the heated carbon monoxide to produce the maximum reduction of the ore. Seven furnaces were constructed and tested before arriving at the one which they considered practical and commercial. This required over two years and an expenditure of $ 102,000. 1(> A vertical section of the furnace is shown in Figure 103, from which the general construction is perfectly obvious. It evi- dently resembles somewhat Stassano's original furnace, and, like his, is started as an ordinary blast furnace. 10 The crucible is 2.25 meters in diameter and 1.5 meters high. The most important point in the construction is the manner in which the electrodes are brought into the melting chamber. As seen from the section, they enter through that portion of the roof of the crucible that does not come in contact with the charge, and pass into the charge at the slope formed by the materials of which it is composed. The electrodes dip into the charge, but not into the melted iron beneath it. 11 Experiments had shown that the brickwork lining around the electrodes was al- ways destroyed if brought in contact with the charge, even when the electrodes were water cooled. The brickwork com- posing the lining of the roof of the melting chamber was cooled by forcing against it, through tuyeres, the compara- tively cool tunnel-head gases. The heat absorbed by these gases is given back to the charge above. A three-phase current is supplied to three electrodes 11 by 22 inches in cross section and 63 inches in length. The water- cooled stuffing boxes through which the electrodes enter the melting chamber are provided with devices to prevent the hot gases under pressure from leaking out around the electrodes. The results of a short run that was made in the presence of Dr. Haanel showed (1) that the furnace operated uniformly and without trouble of any kind for five consecutive days, the electrodes requiring no adjustment whatever; (2) that the energy consumption was remarkably uniform ; (3) that a free 10 For the evolution of the furnace, and dimensions, see Met. and Chem, Eng. 8, 11, (1910). 11 Assar Gronwall, Electrochem. and Met. Ind. 7, 420, (1900). THE ELECTROMETALLURGY OF IRON AND STEEL 251 space was maintained between the charge and the roof of the heating chamber ; (4) that the charge did not jam at the lower contracted neck of the shaft, but moved with regularity into the melting chamber; and (5) that the lining of the roof of the melting chamber was effectively cooled by the circulation of gas. Since the short run witnessed by Dr. Haanel, the furnace has been in continual operation for 85 days, and met all the requirements that indicate a durable furnace. 10 The de- signers of this furnace have contracted to erect three large furnaces for the reduction of iron ores at Sault Ste. Marie, Canada, to be in operation by the middle of 1910. 12 The first electric smelting plant in Canada was under construc- tion at Welland, Ontario, in 1907. 13 It was to consist of a 3000 horse power furnace of the latest type brought out by Heroult. In 1909 an electrothermic plant for reducing iron ore was in existence on the Pitt River at Heroult, Shasta County, California." From the section of this 1500 kilowatt furnace shown in Figure 104, its resemblance to the furnace at Domnarf vet will be evident. A general view is shown in Fig- ure 105. Though this furnace is on a commercial scale, in July, 1910, it was still in the experimental stage, on account 12 Electrochem. and Met. Ind. 7, 535, (1909). is Haanel's Report, 1907, p. 147. " D. A. Lyon, Trans. Am. Electrochem. Soc. 15, 39, (1909). 252 APPLIED ELECTROCHEMISTRY of numerous difficulties that had been encountered. Several changes have been made and it is expected that the furnace will be perfected shortly. When this is accomplished, the Noble Electric Steel Company will build four or five others of a similar type. 15 Pig iron on the Pacific coast brings $23 to $26 a ton, 16 and the cost from this furnace is expected to be $15 a ton, which leaves a good margin of profit. 3. THE ELECTROTHERMIC REFINING OF STEEL While the application of electrothermics to the reduction of pig iron is scarcely an established commercial industry, the case is quite the reverse in steel refining, for a large number of fur- naces for this purpose are in operation in Europe and America. Even in this case, however, the electric furnace cannot compete with the Bessemer or with the open-hearth process for making structural steel. Electric furnace refining is used only to pro- duce very high-class steel for special purposes, 1 for which it is far superior to the crucible process, on account of the greater cheapness and higher quality of the steel produced. 2 The reason for the better quality of the product is that the atmos- phere is neutral, and a much higher temperature can be obtained than by other means, resulting in a more complete removal of impurities, especially gases. Phosphorus and sulphur disap- pear nearly completely, and deoxidation is more complete than that attained by any other means. Another advantage of electric heating is the reliability and certainty of the process. 3 A number of different electric furnaces have been designed for refining steel, and some of the principal ones will now be described. 15 private communication from Professor D. A. Lyon, the manager of the company. 16 Bennie, Trans. Am. Electrochem. Soc. 15, 36, (1909). 1 Haanel's Report, (1904), p. 31; Hibbard, Trans. Am. Electrochem. Soc. 15, 231, (1909). 2 Askenasy, Technische Elektrochemie, p. 56, (1910). 3 Askenasy, Technische Elektrochemie, p. 156, (1910). THE ELECTROMETALLURGY OF IRON AND STEEL 253 FIG. 105. Electric furnace at Heroult, California, for reducing iron ore 254 APPLIED ELECTROCHEMISTRY The furnace used by Stassano at his works in Turin is similar to the one he finally adopted for reducing iron ore 4 (Figures 99 and 100). The charge is heated by radiation from arcs formed between three electrodes placed above the charge and supplied with a three-phase current. This furnace also rotates on an axis slightly inclined to the vertical, in order to mix the charge thoroughly. The lining is magnesite brick. & Starting with scrap and oxidized turnings, about one kilowatt hour is required for one kilogram of finished steel in the 250 horse power furnaces used at Turin. A furnace designed by Charles Albert Keller for steel refin- ing, which was put into industrial use in 1907, is shown in Figures 106 and 107. It consists of a crucible with a conduct- ing bottom for one electrode and a vertical carbon rod for the other. 6 Since carbon must not be brought in contact with the melted iron in refining, the bottom must be made conduct- ing without the use of carbon, and this was accomplished by Keller as follows : Iron bars from 1 to 1| inches in diameter are regularly spaced about one inch apart, and are made fast to a metallic plate at the bottom, covering the entire area on which the bath will rest. Agglomerated magnesia is then rammed, while hot, in between the bars. The whole base is surrounded by a metallic casing for water cooling. Electrical contact is made by the lower plate to which the bars are fastened. The furnace is closed by a cover through which the other electrode passes. After several months' use a hearth constructed in this manner was found to be in as good condition as on the first day. The advantage claimed for this arrangement over a furnace with two vertical electrodes is that the current is more evenly distributed through the charge, and consequently heats it more evenly. Of course, the iron bars are melted at their upper ends where they come in contact with the melted iron to be refined, but the water cooling prevents them from melting for more than a few inches of their length. 4 Trans. Am. Electrochem. Soc. 15, 63, (1909). 6 Trans. Am. Electrochem. Soc. 15, 86, (1909). 6 Trans. Am. Electrochem. Soc. 15, 96, (1909). THE ELECTROMETALLURGY OF IRON AND STEEL 255 I _ 1 > / <* V v ^;oc.a> FIGS. 106 and 107. Keller conducting hearth furnace 256 APPLIED ELECTROCHEMISTRY The Heroult steel refining furnace, 7 as shown in Figure 108, consists of a crucible a with a cover b holding a small chimney c. As the figure shows, it is arranged for tilting, d are car- bon electrodes, which may be moved in a vertical or in a hori- zontal direction. In order to use the furnace for Bessemer- FIG. 108. The Heroult electric steel furnace izing, the tuyeres x are provided. The two electrodes do not quite touch the slag on the surface, so that two arcs are pro- duced. In passing through the bath, the current, of course, divides between the slag and the melted iron in proportion to their conductivities, and as melted iron conducts better than the slag, a larger proportion would flow through the metal than through the slag. The poorest kinds of scrap, high in sulphur and phosphorus, are refined in this furnace. The following table shows the average refining ability of a 2 J-ton furnace at La Praz, Savoy : 7 Electrochem. Ind. 1, 64, (1902) ; U. S. Pat. 707,776. THE ELECTROMETALLURGY OF IRON AND STEEL 257 TABLE 31 PEE CENT Sulphur Phosphorus Manganese Silicon Carbon Scrap charged Finished steel 0.052 0.006 0.150 0.009 0.638 0.254 0.062 0.172 0.211 1.013 For a 5-ton furnace, starting with cold scrap, 600 kilowatt hours are necessary to partially refine one long ton of steel, and 100 more for the finishing slag. For a 15-ton furnace, less power would be required. Figure 109 shows a 15-ton three-phase Heroult furnace at the South Chicago Works of the Illinois Steel Company. The steel to be treated is brought directly from the Bessemer converters, and two refining slags are used in the electric furnace, the first an oxidizing slag to take out the phosphorus, and the second, a deoxidizing slag for removing the sulphur and the gases. 8 Power is supplied to the three electrodes by three transformers, each of 750 kilowatts capacity. Two hundred and forty tons of steel are turned out per day in 16 heats. The electrodes, 2 feet in diameter and 10 feet in length, are the largest ever made in one piece. In cold melting and in continuous work, the consumption of electrode is from 60 to 65 pounds per ton of steel, but when the metal is charged in the melted state, the consumption would be reduced to 10 or 15 pounds per ton of steel. This includes the short ends that cannot be utilized. The linings last from three months to one year, depending on the care with which the furnace is run; the roof suffers most, and generally has to be renewed once a month. The best lining for this furnace is magnesite mixed with basic slag, with tar for a binder. The Paul Girod electric furnace 9 is somewhat similar to the Keller furnace, as seen from Figure 110. One or more elec- 8 Robert Turnbull, Trans. Am. Electrochem. Soc. 15, 139, (1909). 9 Paul Girod, Trans. Am. Electrochem. Soc. 15, 127, (1909). 258 APPLIED ELECTROCHEMISTRY THE ELECTROMETALLURGY OF IRON AND STEEL 259 trodes of like polarity are suspended above the crucible, while the electrode of opposite polarity consists of a number of pieces of soft steel buried in the refractory material of the hearth at its periphery and water cooled at their lower ends. The upper ends come in contact with the bath and are melted to a depth of 2 to 4 inches. About 55 volts are applied to this furnace. For fusing, refining, and finishing a charge of cold scrap in a 2-ton, furnace, about 900 kilowatt hours per metric ton of steel are required, and in an 8 to 10 ton furnace, 700 kil- owatt hours. The electrode con- sumption is 16 to 18 kilograms per metric ton of steel produced in a 2-ton furnace, and 13 to 15 kilograms in an 8 to 10 ton fur- nace. The short ends are in- cluded as having been used. The lining is magnesite or dolomite brick or paste, and lasts 40 to 50 heats without any repairs what- ever. An entirely different class of steel-refining furnaces are those having the melted metal in the form of a ring, forming the sec- ondary of a transformer which is heated by an induced current from a primary coil of copper wire. This type of furnace was patented in 1887 by Colby in the United States and by Ferranti in England. The same principle was applied on a small scale in 1900 by F. A. Kjellin at Gysinge, Sweden, without knowing at the time that it had been patented by others. 10 Kjellin, 10 Kjellin, Trans. Am. Electrochem. Soc. 15, 173, (1909). Fia. 110. The Girod electric steel furnace 260 APPLIED ELECTROCHEMISTRY however, seems to have been the first to carry this idea out on a commercial scale. In 1902 a 225 horse power induction fur- nace was in operation at Gysinge, with an output of 4 metric tons in 24 hours. This furnace had a magnesite lining in FIGS. Ill and 112. Elevation and plan of the Kjellin induction furnace place of silica used in the smaller furnace. A silica lining lasted only about one week, while the magnesite lasted twelve. THE ELECTROMETALLURGY OF IRON AND STEEL 261 Figures 111 and 112 show the principle of the Kjellin furnace. The magnetic circuit C is built up of laminated sheet iron. D is the primary circuit, consisting of a number of turns of insulated copper wire or tubing. The ring-shaped crucible A, for holding the melted metal, is made of refractory material. This furnace cannot be started by placing cold scrap in the crucible because of the low induced electromotive force, but an iron ring must be placed in the crucible and melted down, or the crucible must be filled with melted metal taken from another source. The power consumption of the furnace at Gysinge, starting with cold pig iron and scrap, is about 800 kilowatt hours per metric ton of product. This furnace has been found very satisfactory for making the highest-class steel from pure raw materials. There is a limit to the current that can be sent through the liquid metal, and consequently a limit to the temperature attainable. This is due to a phenomenon first observed by Paul Bary in 1903, 11 to which the name " pinch effect " was given by Hering. 12 This phenomenon is as follows : When a direct or an alternating current passes through a liquid con- ductor, the conductor tends to contract in cross section, forming- a depression, and if the current is large enough, the metal in the trough will separate entirely and break the circuit. This is due, of course, to the attraction the different elements of the current exert on each other. It is most likely to happen at some particular place where the cross section of the ring is smaller than elsewhere, and if any infusible material falls into this depression, it may prevent the reunion of the liquid and cause the charge to .freeze. The largest possible current that could be passed through liquid iron in a trough 2 inches deep and 1 inch wide is about 3300 amperes ; in a trough 4 by 2 inches, 9400 amperes ; and in a trough 6 by 3 inches, 17,000 amperes. 13 Larger currents would cause the metal to separate 11 Northrup, Trans. Am. Electrochem. Soc. 15, 303, (1909). 12 Hering, Trans. Am. Electrochem. Soc. 11, 329, (1907) j 15, 255 and 271, (1909). 18 Trans. Am. Electrochem. Soc. 15, 269, (1909). 262 APPLIED ELECTROCHEMISTRY entirely. When a depression is formed, hydrostatic pressure balances the pressure due to the current, so that this effect is not so likely to give trouble in a deep channel as in a shallow one, nor with a heavy metal as with a light one. It has been found impossible, for instance, to raise aluminum much above its melting point, in a 60 kilowatt induction furnace on account of this effect. 14 The Kjellin furnace is not adapted to working with dephos- phorizing and desulphurizing slags, as the annular ring is not a convenient shape and offers too small a surface to the attack of the slag. 15 A combined induction and resistance furnace was therefore invented by Rodenhauser, known as the Rochling- Rodenhauser furnace for refining Bessemer steel. A plan and an elevation of this furnace are shown in Figures 113 and 114. HH &VQ the two legs of the iron transformer core, surrounded by the primary windings AA. Surrounding the legs of the transformer are the two closed circuits of melted metal, forming together a figure 8, in which currents are induced. BB are two extra primary coils, from which the current is conducted to the metallic plates EE. These are covered by an electrically conducting refractory material, through which the current passes into the main hearth, D. The result is that the main hearth can be made with a much larger cross section than the ring in the original Kjellin furnace, and a good power factor can be obtained in large furnaces without such a low periodicity as was necessary with the original induction furnaces. The mag- nitude of the current from the secondary coils is limited by the carrying capacity of the refractory material 6r, which would be destroyed if too heavily loaded. In refining, the furnace is worked as follows : Fluid steel from the converters is poured into the furnace, and burnt limestone and mill scale are added for forming a basic dephosphorizing slag. This is removed, after the reactions are ended, by tilting the furnace. For mak- ing rails the phosphorus is reduced sufficiently in one opera- tion, but for the highest-class steel it has to be repeated. 14 FitzGerald, Trans. Am. Electrochem. Soc. 15, 278, (1909). 16 Kjellin, Trans. Am. Electrochem. Soc. 15, 175, (1909). THE ELECTROMETALLURGY OF IRON AND STEEL 263 FIGS. 113 and 114. Elevation and plan of the Rochling-Rodenhauser furnace 264 APPLIED ELECTROCHEMISTRY After removing phosphorus, carbon is added in the pure state when carbon steel is to be made, and a new basic slag is formed to remove the sulphur. Rochling-Rodenhauser furnaces are also built for three-phase currents. CHAPTER XIV THE FIXATION OP ATMOSPHERIC NITROGEN I. INTRODUCTION NITROGEN, though chemically an inert element, is of great importance to plant and animal life. It forms 80 per cent by volume of the atmosphere, but it has been impossible until recently to get atmospheric nitrogen in a combined state for use in fertilization or in the chemical industries. This was a problem of the greatest importance, as the nitrogen removed from the soil by crops must be replaced either by adding it in the form of some nitrogen compound or by raising a crop, such as clover, that assimilates the nitrogen of the air by means of a certain kind of bacteroid existing on the root of the plant. Consequently, Chili saltpeter is used in large quantities for fer- tilization, but as this supply is not expected to last later than 1940, 1 the discovery of some other means of supplying the demand became imperative. At present there are three different methods in operation of combining atmospheric nitrogen. The first method con- sists in heating calcium carbide in pure dry nitrogen to about 1000 C., whereby nitrogen is absorbed, forming calcium cyana- mide, according to the reversible reaction : CaC 2 + NCaCN + C. The second method consists in oxidizing nitrogen to nitric oxide in the electric arc and absorbing the oxide in water or in an alkaline solution, and the third and most recent method is the direct synthesis of ammonia from its elements. 1 Edstrom, Trans. Am. Electrochem. Soc. 6, 17, (1904). 265 266 APPLIED ELECTROCHEMISTRY 2. ABSORPTION BY CALCIUM CARBIDE According to Moissan, pure carbide is unaffected by nitro- gen at 1200 C. 1 The discovery that nitrogen is absorbed by commercial calcium carbide and barium carbide was patented in 1895 by Adolph Frank and N. Caro. 2 In the case of barium carbide 30 per cent forms cyanide in place of cyanamide, 3 while in the case of calcium only a trace of cyanide is formed. Since 1895 this reaction has been the subject of a number of investigations. With regard to the temperature required, it has been shown that finely powdered carbide must be heated to from 1000 to 1100 C. to bring about complete transforma- tion to cyanide. At 800 to 900 some nitrogen is absorbed, but the reaction ceases before all the carbide is used up. 4 By the addition of other calcium salts, such as calcium chloride, or, to a less extent, calcium fluoride, complete nitrification can be produced at 700 to 800 C. 5 That the commercial carbide can be completely nitrified at 1100 is due to the presence of calcium oxide. 6 Commercial calcium carbide containing 75 to 80 per cent carbide can be made to take up 85 to 90 per cent of the theoretical amount of nitrogen, forming a black mass of calcium cyanamide, lime, and carbon containing 20 to 23.5 per cent of nitrogen. 3 Pure calcium cyanamide contains 35 per cent nitrogen. The reaction by which it is made is accompanied by a large evolution of heat, which of course is advantageous in its manufacture. According to Caro, this heat is sufficient to cause the reaction to proceed of itself when once started. 7 The system consisting of calcium carbide, calcium cyanamide, carbon, .and nitrogen, is monovariant, that is, for every tem- perature there is a corresponding pressure of the nitrogen at which equilibrium exists. This equilibrium has been meas- 1 C. R. 118, 501, (1894). 2 F ra nk, Z. f. angew. Ch. 19, 835, (1906). 8 Erlwein, Z. f. angew. Ch. p. 533, (1903). 4 Foerster and Jacoby, Z. f. Elektroch. 15, 820, (1909). 5 Bredig, Z. f. Elektroch. 13, 69, (1907). 6 Foerster and Jacoby, Z. f. Elektroch. 13, 101, (1907). * N. Caro, Z. f. angew. Ch. 22, 1178, (1909). THE FIXATION OF ATMOSPHERIC NITROGEN 267 ured between 1050 C. and 1450 C., and the results are given in the plot in Figure 11 5. 8 If the initial pressure of nitrogen lies in the region above the line, absorption of nitrogen takes place, while if below, any calcium cyanamide present would 50 40 30 20 10 4B 1000 1100 1400 1500 1200 1300 v TEMPERATURE FIG. 115. Plot showing pressures and temperatures at which equilibrium of the reaction CaC 2 +N 2 ^:CaCN2 + C exists. Pressures are in centimeters of mer- cury ; temperatures in centigrade degrees. decompose until the nitrogen produced brings the pressure up to that corresponding to equilibrium, or until all of the cyan- amide is used up. The velocity of absorption of nitrogen is proportional to its pressure, 9 assuming other conditions constant. At a constant 8 Thompson and Lombard, Proc. Am. Acad. 46, 247, (1910) ; Met. and Chem. Eng. 8, 617, (1910). During proof reading the experiments of Le Blanc and Eschmann, with results different from those above, appeared ; see Z. f. Elek- troch. 17, 20, (1911). They find that the pressure depends on the nitrogen con- tent of the solid phase as well as on the temperature. Bredig, Fraenkel, and Wilke, Z. f. Elektroch. 13, 605, (1907). 268 APPLIED ELECTROCHEMISTRY temperature, with a constant surface of carbide exposed, and a given amount of nitrogen in a given volume, this law is expressed by the differential equation : where p is the pressure, t the temperature, and k is a constant. Integrated this becomes 0.43 1 Pi where p 2 and p l are the pressures at the beginning and end respectively of the time interval t. Calcium cyanamide acts in some cases as the calcium salt of cyanamide : Ca = N C s N, and in others as the calcium salt of the diimide With superheated steam the nitrogen is changed to ammonia according to the reaction 2 CaCN 2 + 3 H 2 = CaCO 3 + 2 NH 3 with a yield 99 per cent. 10 Dicyandiamid, a compound con- taining 66 per cent nitrogen, can be made by treating calcium cyanamide with water. It has the appearance of ammonium chloride, and is probably formed by the following reaction : 2 CaCN 2 + 4 H 2 = 2 Ca(OH) 2 + (CNNH 2 ) 2 . n Calcium cyanide can be made from technical calcium cyana- mide by melting with a suitable flux, such as sodium chloride, according to the following reversible reaction : This use of calcium cyanamide is second in importance only to its direct application as a fertilizer. 2 According to Frank 2 one horse power year can produce enough carbide to absorb 772 kilograms of nitrogen, though the value actually realized amounts to only 300 to 330 kilo- 10 Erlwein, Z. f. Elektroch. 12, 551, (1906). 11 Z. f. angew. Ch. p. 520, (1903). THE FIXATION OF ATMOSPHERIC NITROGEN 269 grams. According to a later statement by Caro, 7 3 horse power years is more than sufficient to absorb one metric ton of nitro- gen, including the manufacture of the carbide and all the other power required in the factory for the grinding and moving apparatus, the Linde machines for liquefying air, and so forth. Thus a factory with 12,000 horse power produces yearly 20,000 metric tons of calcium cyanamide containing 20 per cent nitro- gen, corresponding to 4000 metric tons of nitrogen. It is in- teresting to compare these data with the power required to produce the corresponding amount of calcium carbide. An average yield of carbide has been shown above to be 5.5 kilos of 80 per cent carbide per kilowatt day, corresponding to 1500 kilos per horse power year. 376 kilograms of nitrogen would have to be absorbed by this amount of carbide in order that the product should contain 20 per cent nitrogen. This is a little above the value 300 to 330 actually obtained as given by Frank. If the statement of Caro is correct, and carbide is produced with the efficiency assumed above, it means that 90 per cent of the power in a cyanamide factory is used for producing the carbide alone. Nitrogen is obtained by the Linde process or by removing the oxygen with hot copper. It must be free from oxygen, for this would produce carbon monoxide, which decomposes both carbide and cyanamide. 7 Caro states that moisture must be also absent, though Bredig, Fraenkel, and Wilke's 9 experiments showed that when the nitrogen was saturated with water vapor at 22, a little more nitrogen was absorbed than when dry. Besides lime and carbon, there are impurities in technical cyanamide, consisting of nitrogen compounds, such as urea, guanidine, and calcium carbamate. In fresh samples these impurities are small in quantity, but increase on standing or by the presence of water vapor. All of these substances are easily assimilated by plants. 7 The manufacture of calcium cyanamide was begun on a large scale in 1905 at Piano d'Orta, Italy, 10 and in 1908 there were 11 factories in Europe making this substance. 12 Norway and i2Min. Ind. 17, 105, (1908). 270 APPLIED ELECTROCHEMISTRY Sweden are unusually favorable localities for the nitrogen industry on account of the large amount of cheap water power. Recent estimates on power in these countries are as follows : 13 AVAILABLE DEVELOPED Sweden 4,000,000 h. p. 400,000 h. p Norway . 5 000 000 h. p. 500 000 h p The figures under " developed " refer to plants in operation or under construction. There is a 20,000 horse power plant for the production of cyanamide and calcium carbide at Odda, Norway, having a capacity of 32,000 short tons of carbide and 12,500 tons of cyanamide per year. The nitrogen, which must not contain over 0.4 per cent oxygen, is obtained by the Linde process. The furnaces in which the carbide is heated with nitrogen are charged with about 700 pounds and produce 2000 pounds of cyanamide containing 20 per cent nitrogen per week. In 1909 this industry was introduced on this side of the Atlantic by the American Cyanamide Company, which owns the exclusive rights for manufacturing nitrolime in this country. A factory is now in operation at Niagara Falls, Ontario. 14 The product is to contain 12 to 15 per cent nitrogen, 10 per cent carbon, and 25 per cent calcium sulphate. Free lime is to be eliminated as is demanded by American trade. 3. THE OXIDATION OF NITROGEN Priestley l was the first to observe that electric sparks in air produced an acid, though he mistook it for carbonic acid. Later Cavendish 2 repeated the experiments and showed the true nature of the acid produced, which is now known to be a mixture of nitrous and nitric acids. From the time of Caven- 13 Electrochem. and Met. Ind. 7, 212 and 360, (1909). 14 Met. and Chem. Eng. 8, 227, (1910). 1 Experiments and Observations on Different Kinds of Air, 4, 286. Preface dated 1779. Also Ostwald, Elektrocheinie, p. 11. a Phil. Trans. 75, 372-384, (1797). Also Alembic Club Reprints, No. 3, p. 39. THE FIXATION OF ATMOSPHERIC NITROGEN 271 dish until within the last twenty years nothing of importance was done toward explaining this phenomenon. Since 1890, however, it has received considerable attention, so that now, principally due to the work of Nernst and Haber, the conditions under which the reaction N 2 + O 2 ^ 2 NO takes place are well known. Nernst and his assistants have measured the thermal equilib- rium concentrations of nitrogen, oxygen, and nitric oxide at different temperatures with the results in Table 32. 3 TABLE 32 Per cent by Volume of Nitric Oxide in the Equilibrium Mixture formed from Air DEGREES ABSOLUTE PER CENT NO OBSERVER 1811 Observed 0.37 Computed 0.35 Nernst 1877 0.42 0.43 Jellinek 2023 0.52 to 0.80 0.64 Jellinek 2033 0.64 0.67 Nernst 2195 0.97 0.98 Nernst 2580 2.05 2.02 Nernst-Finckh 2675 2.23 2.35 Nernst-Finckh The values in the third column were computed by the Van't Hoff equation, with Berthelot's value of 21,600 calories for the heat of the reaction. These experiments show that at the temperatures given the velocity of decomposition is so low that the gas can be cooled without decomposition of the nitric oxide already formed. The free energy of the reaction is given by the equation 4 = Q -ET log- + 2.45 T, in which Q = 21,600 calories. By means of this equation the per cent of nitric oxide corresponding to the equilibrium at any temperature can be computed by placing the right-hand Z. f. anorg. Ch. 49, 213, (1906). * Haber, Thermodynamics of Technical Gas Reactions, p. 105, (1908). 272 APPLIED ELECTROCHEMISTRY side equal to zero, which is the equilibrium condition. The experiments of Finckh were carried out by exploding air mixed with detonating gas ; the others by drawing air through plati- num or iridium tubes heated electrically. The good agreement between the calculated and observed values shows that at least in these experiments the nitric oxide formed is due only to the high temperature, as the concentration is that required by thermodynamics. This reaction is bimolecular between 650 C. and 1750 C., 5 that is to say, it should be written N 2 + O 2 = 2 NO. Le Blanc and Niiranen, however, have found that above 3000 C. the reaction is monomolecular. 6 Tables 33 and 34 give the veloci- ties of the reaction in both directions at different temperatures. 6 TABLE 33 Time in Minutes necessary to decompose Pure Nitric Oxide at Atmospheric Pres- sure, Half into Nitrogen and Oxygen DEGREES ABSOLUTE TIME IN MINUTES DEGREES ABSOLUTE TIME IN MINUTES 900 7.35 10 3 2100 1.21 10- 3 1100 5.80 10 2 2300 8.40 10- 5 1300 4.43 10 1 2500 5.76 10- 6 1500 3.30 2700 3.92 10- 7 1700 2.47 10- 1 2900 3.35 10~ 8 1900 1.47 lO- 2 3100 2.25 10- 9 TABLE 34 Time required to produce from Air One Half the Possible Amount of Nitric Oxide DEGREES ABSOLUTE TIME IN MINUTES DEGREES ABSOLUTE TIME IN MINUTES 1500 1.81 10 8 2500 1.77 10~ 4 1700 5.90 10 1 2700 8.75 10- 6 1900 2.08 2900 5.75 10- 7 2100 8.43 10- 2 3100 3.10 10- 8 2300 3.75 10-8 5 Jellinek, Z. f. anorg. Ch. 49, 229, (1906). 6 Z. f. Elektroch. 13, 303, (1907). THE FIXATION OF ATMOSPHERIC NITROGEN 273 From these results it would appear that the best yield of nitric oxide would be obtained by heating the gas to the highest temperature from which it could be chilled so suddenly that decomposition would not take place. It has been shown, however, that nitric oxide can be produced by the silent dis- charge of electricity where there is very little elevation of temperature. 7 This fact suggested to Haber and Koenig 8 the possibility of obtaining better yields by using a comparatively cool arc, which could be realized by inclosing it in a tube surrounded by water. Below 3000 C. any oxide produced by the impact of electrons would not be decomposed rapidly by the heat even if the concentration due to the electrical effect were greater than that due to the thermal. In fact they found that by using a cooled arc and by reducing the pressure to the most favorable value of 100 millimeters, concentrations of nitric oxide were obtained which could be explained thermally only on the assumption that the thermal equilibrium correspond- ing to over 4000 absolute had been obtained and that the gas had been chilled suddenly enough to preserve it. Such a high TABLE 35 Concentrations of Nitric Oxide obtained at 100 mm. Pressure by an Arc inclosed in a Cooled Tube INITIAL GAS MIXTURE IN THERMODYNAMICALLY COM- PER CENT BY VOL. 7>NO NO CONTENT PUTED TEMP. ABS. i i IN PER CENT _) O 2 ft N^2 -y o, N, Haber Nernst 20.9 79.1 0.284 9.8 4365 4334 48.9 51.1 0.337 14.4 4686 4650 44.4 55.6 0.337 14.3 4686 4650 75.0 25.0 0.357 12.77 4805 4767 81.7 18.3 0.397 12.1 5042 5000 7 Warburg and Leithauser, Ann. d. Phys. (4) 20, 743, (1906), and 23, 209, (1907). 8 Z. f. Elektroch. 13, 725, (1907). 274 APPLIED ELECTROCHEMISTRY temperature in their arc seemed impossible ; consequently the oxide must have been produced directly by the impact of ions. Table 35 gives the concentrations of nitric oxide obtained with the temperature corresponding, on the improbable assumption that this concentration corresponds to a thermal and not to an electrical equilibrium. The temperatures were computed both by Haber's formula given above and by the Van't Hoff formula as used by Nernst. In later experiments as high as 17.8 per cent nitric oxide was obtained. 9 It was further found that the same concentration is obtained under similar conditions from either nitric oxide or from air and oxygen, showing that we have in this case an electrical equilibrium. If the temperature is too high, the electrical equilibrium is obliterated by the thermal. On the other hand, the electrical energy necessary to produce ioniza- tion increases considerably when the temperature falls below white heat. There will therefore be a most favorable region of temperature within which the nitric oxide produced by the impact of ions will not be decomposed and when too much electrical energy is not required for ionization. 8 It would, therefore, seem that the best way to try to obtain better re- sults is to employ a cool arc rather than by attempting to heat to a higher temperature and chill more suddenly. The energy efficiency was not determined in these experi- ments. In later ones, 10 with a cooled arc, the efficiency, when the concentration of the nitric acid obtained was 3.4 per cent, was 57 grams of nitric acid per kilowatt hour, or 500 kilo- grams per kilowatt year of 365 x 24 hours. With a cooled arc and a direct current, Holweg and Koenig n obtained nitric acid at a concentration of 2.5 per cent and an efficiency corresponding to 80 grams of nitric acid per kilowatt hour, the most favorable energy efficiency ever reached. Increas- ing the pressure above atmospheric does not increase this efficiency. 12 Z. f. Elektroch. 14, 689, (1908). 10 Z. f. Elektrock. 16, 795, (1910). " Z. f. Elektroch. 16, 809, (1910). 12 Haber and Holweg, Z. f. Elektroch. 16, 810, (1910). * THE FIXATION OF ATMOSPHERIC NITROGEN 275 On cooling down, the colorless nitric oxide changes to the brown dioxide of nitrogen, since the reversible reaction NO + \ r 2 ^ N0 2 is displaced from left to right on cooling. Table 36 shows how the dissociation of nitrogen dioxide is affected by the temperature : 13 TABLE 36 DEGREES CENTIGRADE PRESSURE IN CENTIMETERS PER CENT OF NO 2 DECOMPOSED 130 71.85 184 75.46 5.0 279 73.72 13.0 494 74.25 56.5 620 76.00 100.0 It will be interesting to compute from a purely thermal standpoint the energy necessary to produce nitric acid and to compare this result with those actually found by different ex- perimenters. Assuming the temperature of the high tension arc to be 4200 C., the calculation is as follows. 14 From the equation given above at this temperature =0.29, and if the original mixture is air, the final composition is : NO O 2 N 2 10 per cent 16 per cent 74 per cent Ten moles of nitric oxide with air and water yield 630 grams of nitric acid. Therefore, in order to get this amount of acid, 100 moles must be heated to 4200 C., besides which 10 x 21,600 calories must be supplied for the reaction. Assuming the spe- 18 Nernst, Theoretische Chemie, p. 455, 6th ed. See also Bodenstein and Kata- yama, Z. f. Elektroch. 15, 244, (1909). 14 Haber, Thermodynamics of Technical Gas Reactions, p. 268. 276 APPLIED ELECTROCHEMISTRY cific heat of the permanent gases to be 6.8 -f 0.0006 calories per mole, the total energy will be : 100 (6.8 + 0.0006 x 4200) 4200 + 216,000 = 4,130,000 calories. This corresponds to 4.71 kilowatt hours for 630 grams of nitric acid, or 134 grams per kilowatt hour. If the arc were 1000 lower, the result would be 93.5 grams per kilowatt hour. The results obtained with a cooled arc are not due to ther- mal equilibrium, and of course have no relation to this calcula- tion. Unless special precautions were taken to use a cooled arc, the results may be assumed to be due to thermal and not to electrical causes. This is the case in the following examples. Lord Rayleigh 15 obtained an absorption of 21 liters an hour with 0.8 kilowatt, using a mixture of 9 parts of air and 11 of oxygen. This corresponds to 46 grams of pure nitric acid per kilowatt hour, assuming the gas was measured at 20 C. and at atmospheric pressure. McDougall and Howies 16 with an ar- rangement similar to that of Lord Rayleigh obtained 33.5 grams of nitric acid per kilowatt hour. McDougall and Howies were the first to make a small experimental plant for the production of nitric acid from the air. 17 It seems not to have got beyond the experimental stage, however. The first 18 attempt to carry out the oxidation of nitrogen on a commercial scale was that of the Atmospheric Products Com- pany at Niagara Falls, using the patents of Bradley and Love- joy. Their first apparatus 19 was similar to that of McDougall and Howies and consisted in a number of small compartments in which an arc was formed between electrodes in the form of a hook at the points nearest together, as shown in Figure 116. The arc then ran along the electrodes, thereby becoming longer, until it went out, whereupon the arc was formed again. This i 5 Journ. Chem. Soc. 71, 181, (1897). , 16 Memoirs and Proceedings of the Manchester Literary and Phil. Soc. (IV) 44, 1900, No. 13. 17 Huber, Zur Stickstoff Frage, p. 41, Bern, (1908). 18 Donath and Frenzel, Die Technische Ausinetzung des Atmospharischen Stickstoffes, p. 126, (1907). U. S. Pat. 709,867, (1902). THE FIXATION OF ATMOSPHERIC NITROGEN 277 arrangement was supplanted by a single apparatus, shown in Figures 117 and 118, in which 6900 arcs were formed per second. 20 This consisted in an iron cylinder 5 feet high, 4 feet in diameter, in the center of which was a rotating shaft carry- ing a series of radial arms, the ends of which were tipped with FIG. 116. First apparatus of Bradley aud Lovejoy platinum. Six rows of 23 inlet wires projected through the cylinder and terminated in a platinum hook. As the radial arms rotated, their platinum tips passed the hooks on the inlet wires, coming within one millimeter of touching at the nearest point. An arc was formed which was drawn out from 4 to 6 inches before going out. The arms were so arranged that the 20 J. W. Richards, Electroch. Ind. 1, 20, (1902) ; U. S. Pat. 709,868, (1902). 278 APPLIED ELECTROCHEMISTRY arcs between them and the inlet wires were formed successively rather than simultaneously. The central shaft made 500 rota- tions per minute. Each inlet wire had in series with it an in- duction coil 12 inches long and 5 inches in diameter, wound with very fine wire and immersed in oil. The self-induction of the coil caused the spark to be drawn out to a greater length than would be possible without induction. A direct current FIG. 117. Vertical section of final apparatus of Bradley and Lovejoy generator was especially designed for this plant, giving 8000 volts and 0.75 ampere. Air passed in at the rate of 11.3 cubic meters per second and came out of the cylinder containing 2.5 per cent nitric oxide. 21 The yield is said to have been one pound of acid per 7 horse power hours, or 87 grams per kilo- watt hour. The process was not successful, however, and the company was forced to give up the experiments in 1904. 21 Haber, Z. f. Elektroch. 9, 381, (1903). THE FIXATION OF ATMOSPHERIC NITROGEN 279 Though the yield compared favorably with the calculations given above, the . apparatus was very complicated and subject to considerable wear. The iron drum corroded rapidly in spite of the inside coating of asphalt paint. 18 The first successful process for oxidizing nitrogen on a com- mercial scale is that of Birkeland and Eyde. A factory for carrying it out was started at Notodden, Norway, in May, 1905. ^ The high voltage flame is formed between two electrodes con- sisting of water-cooled copper tubes 1.5 centimeters in diameter FIG. 118. Horizontal section of final apparatus of Bradley and Lovejoy with 0.8 centimeter between the ends. An alternating current of 50 cycles per second is supplied to the electrodes at 5000 volts. In order to spread the flame over a large area an electromagnet is placed at right angles to the electrodes so that the terminals lie between the poles of the magnet. The voltage is sufficiently high to cause the flame to form of itself between the electrodes at their nearest points, whereupon the magnetic field causes the ends of the flame to travel along the electrodes until the current is reversed. A new flame is then started on the other side of the electrodes. When the furnace 22 Birkeland, Trans. Faraday Soc. 2, 98, (1906). 280 APPLIED ELECTROCHEMISTRY is running properly a flame is formed at each reversal of the current every ^ of a second, though if the distance between FIG. 119. Electric disc in the furnace of Birkeland and Eyde the electrodes is too short or the magnetic field too strong, several hundred flames may be started during one period. The magnetic field is J V 4000 to 5000 lines per square centi- meter at the center. The result of this combination is an electric disk flame, as shown in Figure 119. This is in- closed in a narrow iron furnace lined with fire brick, form- ing a chamber from 5 to 15 centimeters wide, shown in Figure 120. Air 120. - Vertical section of furnace of Birkeland I* 8808 in tlir U S h and Eyde the walls and leaves THE FIXATION OF ATMOSPHERIC NITROGEN 281 the furnace at a temperature between 600 and 700 C., con- taining one per cent of nitric oxide. From the furnace the gases pass through a steam boiler in which they are cooled to 200 C., and then through a cooling apparatus in which their temperature is reduced to 50 C. They then enter oxidation chambers with acid proof lining, where the reaction NO + $ O 2 = NO 2 is completed. The next step is to absorb the nitrogen dioxide. This is done in two sets of five stone towers whose inside dimensions are 2 x 2 x 10 meters. The first four towers are filled with broken quartz over which water trickles. The fifth tower is filled with brick, and the absorbing liquid is milk of lime, giving a mixture of calcium nitrate and nitrite. Nitric acid is formed in the first four towers with concentrations as follows : FIRST SECOND THIRD FOURTH 50 % HXO 3 25 % HNO 3 15% HNO 3 5% HNO 3 The liquid from the fourth tower is raised by compressed air to the top of the third, that from the third to the top of the second, and so on until fifty per cent nitric acid is formed. Some of this acid is used to decompose the nitrate-nitrite mixture from the fifth tower. The nitric oxide thereby evolved is sent into the absorbing system again. About 97 per cent of the entire quantity of nitrous gases passed through the absorbing system is absorbed. 23 The resulting solution of calcium nitrate and the rest of the stored-up acid is treated in another set of tanks with lime, producing neutral calcium nitrate. This is evaporated in iron by the steam from the boilers above mentioned till a boiling point of 145 C. is reached, corresponding to 75 or 80 per cent nitrate and containing 13.5 per cent of nitrogen. This is poured into iron drums of 200 liters capacity, where it solidifies. Another method is to crystallize from a boiling point of 120 C. This yields calcium nitrate with four molecules of water. In 1906 at the Notodden Saltpeter Manufactory there were three 500-kilowatt furnaces in constant activity. The volume 28 Eyde, Electrochem. and Met. Ind. 7, 304, (1909). 282 APPLIED ELECTROCHEMISTRY of air treated was 75000 liters per minute. The yield was about 500 kilograms of pure nitric acid per kilowatt year, or 57 grams per kilowatt hour. In place of the smaller furnaces those now used absorb 1600 kilowatts, of which 35 are now in operation at Notodden, 8 in series. The disk flame has a diameter of 2 meters and a thickness of 10 centimeters. 24 During the year 1908 the profits of the Notodden factory were 25 per cent of the total receipts, amounting to 500,000 krone, or $ 135,000. M The company using the Birkeland-Eyde process has combined with the Badische Anilin und Sodafabrik, which has developed another furnace, described below, so that the re- sults of a factory under construction at Notodden in 1909 will decide which furnace will be the one for the final large plant. 25 Up to February, 1909, 16,000,000 had been invested at Notodden and Svalgfos and on the rivers Rjukan and Vamma. By the end of 1910 these plants will be completed and the investment will amount to fl^OOO^OO. 25 The furnace of the Badische Anilin und Sodafabrik of Ludwigshafen, Germany, was invented in 1905 by Schonherr and Hessberger. 26 An alternating current arc is very easily extinguished, especially if air is blown across it. The principle underlying this furnace is that an alternating current arc loses its unstable character and becomes as quiet as a candle if a cur- rent of air is passed around it in a helical path. With this method of air circulation the arc may be included in a metallic tube without risk of its coming in contact with the sides of the tube. A cross section of the apparatus is shown in Figure 121. It consists of a number of concentric vertical iron tubes. The electrode at the bottom is an iron rod adjustable within a water- cooled copper cylinder. The iron is slowly eaten away, and is fed in at about the rate of one electrode in three months. The electrode Z is for starting the arc by bringing it in contact with 24 Birkeland, Electrochem. and Met. Ind. 7, 305, (1909). 25 Eyde, Z. f. Elektroch, 15, 146, (1909). 2^ Electrochem. and Met. Ind. 7, 245, (1909) ; Trans. Am. Electrochem. Soc. 16, 131, (1909). FIXATION OF ATMOSPHERIC NITROGEN 283 E. There is of course an induction coil in series with the arc to make it steady \ and prevent the current from being too large on starting. When Z is drawn back the arc is formed between E and * the walls of the tube. The air then | drives it up along the tube until it | reaches the other water-cooled end, JT, within which the arc terminates. 6r r 2 6r 2 , and 6r 3 are peep holes for observing ! , the ends of the arc. In the 600 horse 1 .- power furnaces at Kristianssand, Nor- bj> way, the arc is 5 meters long, and 7 8 | meters in the 1000 horse power fur- H ' naces. The circulation of the air is g evident from the figure. The plant at Kristianssand, the fur- 3 nace room of which is shown in Figure 122, has been in operation since the autumn of 1907. Three-phase currents 4 are used, and the furnaces are connected in star. The power factor varies between H 0.93 and 0.96. It is estimated that 3 per cent of the power is used in the formation of nitric oxide, 40 per cent is recovered in the form of hot water, 17 per cent is lost by radiation, 30 per cent is used in the steam boiler, and 10 per cent is removed by water cooling after the erases have passed the steam boiler. O T The nitric oxide is absorbed by milk of lime. The final product is calcium nitrite containing 18 per cent nitrogen. The yield per kilowatt hour is not given. A third process for the fixation of at- mospheric nitrogen, invented by H. and G. Pauling, is carried out near Inns- 284 APPLIED ELECTROCHEMISTRY bruck, Tirol, by the " Salpetersaure-Industrie-Gesellschaft." 27 The arcs are produced between curved electrodes, as shown in Figure 123. The arc is lighted where the electrodes are near- est together, is blown upwards by the hot air rising between FIG. 122. Furnace room at Kristianssaud the electrodes, and is broken every half period of the alternating current. Another arc is then formed, and so on. In Figure 123 c represents two thin adjustable blades for starting the arc. Air is blown in through the tube e. The electrodes are iron pipes, water-cooled and separated by about 4 centimeters at , 27 Electrochem. and Met. Ind. 7, 430, (1909). THE FIXATION OF ATMOSPHERIC NITROGEN 285 their nearest point. Their life is about 200 hours. With a 400 kilowatt furnace of 4000 volts the length of the flame is about one meter. Cooling is produced by passing cold air into the upper part of the flame from the side. The concentration of the nitric oxide is about 1.5 per cent. The furnaces used have two arcs in series. Six hundred cubic meters of air per hour pass through the fur- nace, excluding the cooling air. The yield is 60 grams of nitric acid per kilo- watt hour. At pres- * FIG. 123. Electrodes in lurnace of ent there are 24 fur- H. and G. Pauling naces in operation at Innsbruck, having a capacity of 15,000 horse power. The products are nitric acid and sodium nitrite. Two other plants for carrying out this process, each of 10,000 horse power, are in course of erection, one in southern France and the other in northern Italy. A number of other furnaces for the oxidation of nitrogen have been invented, but their descriptions are omitted here because they are not in operation on a commercial scale. 4. THE SYNTHESIS OF AMMONIA The third method of fixing nitrogen, that has just recently been taken up by the Badische Anilin und Sodafabrik, 1 is to make it combine directly with hydrogen to form ammonia, ac- cording to the reversible reaction : This reaction takes place from left to right with the evolution of about 12,000 calories, 2 so that the quantity of ammonia gas in the equilibrium mixture decreases as the temperature rises. 1 Haber, Z. f. Elektroch. 16, 242, (1910). 2 Landolt and Bernstein's Tables, 3d ed. p. 427. 286 APPLIED ELECTROCHEMISTRY The velocity of the reaction, on the other hand, of course in- creases with the temperature, but does not reach a value that adjusts the equilibrium rapidly below a temperature of 750 C. 3 The composition of the equilibrium mixtures for different tem- peratures and two different pressures, when the free hydrogen and nitrogen are present in the same proportion as in ammonia, is given in Table 37. 4 TABLE 37 DEGREES PRESSURE IN VOL. PER CENT PRESSURE IN VOL. PER CENT CENTIGRADE ATMOSPHERES NH, ATMOSPHERES NH 3 700 30 0.654 1 0.0221 801 30 0.344 1 0.0116 901 30 0.207 1 0.00692 974 30 0.144 to 0.152 - 1 0.0048 to 0.0051 It is evident from this table that unless some catalytic agent can be found that would give the reaction high velocity at a temperature considerably below 750, very little ammonia could be obtained at atmospheric pressure. Since, however, there is a decrease in volume when ammonia is formed from an equiva- lent amount of nitrogen and hydrogen, there must be an in- crease in the relative amount of ammonia in an equilibrium mixture when the pressure is increased. It is evident from the table that the volume per cent of ammonia in such a mixture is directly proportional to the pressure, as long as the relative amounts of free hydrogen and nitrogen are kept constant. Jost 5 has obtained somewhat lower values for the amount of ammonia in the equilibrium mixture. Table 38 gives his results obtained at a total pressure of one atmosphere, and those of Haber taken from the table above for comparison. Haber's results at one atmosphere are in good agreement with the values calculated from his results at 30 atmospheres, and therefore are more reliable than Jost's. 8 Haber, Thermodynamics of Technical Gas Reactions, p. 202, (1908). * Haber and Le Rossignol, Z. f. Elektroch. 14, 193, (1908). 5 Z. f. Elektroch. 14, 373, (1908). THE FIXATION OF ATMOSPHERIC NITROGEN TABLE 38 287 VOLUME PER CENT NH 8 T* r* Haber Jost 700 0.0221 0.018 800 0.0116 0.0090 900 0.0069 0.0050 974 0.0048 to 0.0051 0.0035 Haber lias subsequently developed this process further and showed in a lecture 1 a small apparatus working at 185 atmos- pheres that produced hourly 90 grams of liquid ammonia. In the earlier experiments finely divided iron on asbestos was used as a catalyzer, but in these later experiments, uranium was substituted for iron. This method is said to require compara- tively little power, and will therefore not be confined to places where cheap water power is available. No numerical values of the efficiency of this method, however, are given. 5. CONCLUSION Having described the three general methods of fixing atmos- pheric nitrogen now in operation, it will be interesting to com- pare the actual amounts of nitrogen fixed for a given amount of power by the three methods. This is possible only for the absorption by carbide and the direct oxidation. Since 12,000 horse power or 8850 kilowatts can fix 4,000,000 kilograms of nitrogen l per year as calcium cyanamide, one kilo- watt hour corresponds to 51.6 grams of nitrogen. The yield by the Birkeland-Eyde process is about 57.1 grams of pure nitric acid per kilowatt hour, 2 corresponding to 12.7 grams of nitrogen. The cyanamide process therefore fixes about four times as much nitrogen as the direct oxidation for the same ex- penditure of power. i Frank, Z. f. angew. Ch. 19, 835, (1906). 2 Birkeland, Trans. Faraday Soc. 2, 98, (1906); Haber, Z. f. Elektroch. 10, 551, (1906). CHAPTER XV THE PRODUCTION OF OZONE 1. GENERAL DISCUSSION IN 1785 Van Marum observed that oxygen through which an electric spark had passed had a peculiar odor, and that it at once tarnished a bright surface of mercury. 1 Nothing was done to throw light on this phenomenon until 1840, when it was investigated by Schonbein. He had observed for a num- ber of years previously that during the electrolysis of aqueous solutions an odor is produced in the gas evolved at the anode similar to that resulting from the discharge of electricity from points. 2 He described a number of the properties of this sub- stance, and suggested the name ozone, from 6'&>i>, meaning smelling. For many years the chemical nature of this oxidiz- ing principle was unknown, but it was found eventually, after a great number of investigations, to be simply condensed oxygen with the formula O 3 . The formation of ozone from oxygen is an endothermic reaction. The heat absorbed in the production of one mole of ozone, as determined by different investigators, is given in the following table : 3 Berthelot, indirect, 1876 29,800 calories Mulder and v. d. Meulen, indirect, 1883 . . . 33,700 calories v. d. Meulen, indirect, 1882 32,800 calories v. d. Meulen, direct, 1883 36,500 calories Jahn, direct, 1908 . 34,100 calories 1 Roscoe and Schorlemmer, Treatise on Chemistry, 1, 256, (1905). 2 Pogg. Ann. 50, 616, (1840). 8 Stephan Jahn, Z. f. anorg. Ch. 48, 260, (1905). 288 THE PRODUCTION OF OZONE 289 Since heat is absorbed in the production of ozone, thermo- dynamics requires that the equilibrium existing in a mixture of oxygen and ozone be displaced in the direction of a greater ozone concentration by an increase in the temperature of the mixture. In order to prove this experimentally, it is necessary to heat the oxygen to a temperature high enough to produce a measurable quantity of ozone, and then, by cooling suddenly, to prevent the decomposition of the ozone formed. This has been done by blowing air or oxygen against a hot pencil, such as is used in a Nernst lamp, 4 and also by dipping a hot Nernst pencil, or hot platinum, in liquid air. 5 The free energy decrease which accompanies the decomposi- tion of ozone into oxygen has been determined from potential measurements. 6 At C. the potential of the cell O 3 | electro- lyte H 2 equals 1.90 volts, and that of the cell O 2 | electrolyte H 2 equals 1.25 volts. The reactions which take place in these two cells, with the corresponding free energy changes, are therefore given by the following equations : 2 O 3 + 2 H 2 = 2 O 2 + 2 H 2 O + 4 F x 1.90 joules, O 2 + 2 H 2 = 2 H 2 O + 4 F x 1.25 joules, where F is the electrochemical equivalent. The difference between these two equations gives : 2 O 3 = 3 O 2 + 4 F x 0.65, or O 3 = f O 2 + 30,000 calories. From this result the following equilibrium concentrations at high temperatures may be calculated : Temperature on absolute scale . . 1000 1400 1800 2200 Pres. ozone in atmospheres, in equi- librium with oxygen at one atmosphere 0.000029 0.0032 0.038 0.18 (0.007) (0.03) The above results are only approximate, for the very divergent values inclosed in parentheses are within the experimental error. 4 Fischer and Marx, B. B. 40, 443, (1907). 5 Fischer and Braemer, B. B. 39, 996, (1906). 6 Stephan Jahn, Z. f. anorg. Ch. 60, 332, (1908). 290 APPLIED ELECTROCHEMISTRY It will be seen from these results that ozone, in the concen- trations ordinarily prepared, amounting to several per cent by volume, is in a state of unstable equilibrium, and it conse- quently decomposes slowly on standing. This reaction is lii molecular ; 7 that is, 'J' dn = -kn* dt, where n is the number of moles per cubic centimeter, k is a constant, and t is the time. The velocity of this reaction is given in Table 39. ft is the number of grams of ozone in one liter that would decompose per minute if its initial concentra- tion were one gram per liter. TABLE 39 TEMPERATURE 16 0.0000492 100 0.157 126.9 1.77 At 16 one per cent of pure ozone would decompose in 1.7 hours, and 50 per cent in 167 hours. These values apply to ozone in contact with concentrated sulphuric acid, over which the pressure of water vapor is 0.0021 millimeter of mercury. If the pressure of water vapor is 0.154 millimeter, the velocity of decomposition at lOO^is found to be 22 per cent greater. The decomposition of ozone takes place in steps, the reaction whose velocity is measured being O + 3 = 2 2 . 8 Ozone may be produced by the action of ultra-violet light, and of the silent discharge of electricity on oxygen ; by heat- ing and suddenly chilling oxygen, and by electrolysis. While the silent electric discharge is the only method used commer- cially for the manufacture of ozone, it will be interesting to 7 Warburg, Ann. d. Phys. 9, 1286, (1902), and 13, 1080, (1904). 8 Jahn, Z. f. anorg. Ch. 48, 260, (1005). THE PRODUCTION OF OZONE 291 compare the yield per kilowatt hour attained by the silent dis- charge with some of the other methods. By blowing air against a hot Nernst pencil, the yield was found to be one gram 4 per kilowatt hour ; and by dipping hot bodies in liquid air, about 3.5 grams. 5 The concentration of the ozone in bot'^i cases was less than three per cent. By electrolyzing solutions of sulphuric acid of specific gravity between 1.075 and 1.1 with a water-cooled platinum anode, as high as 17 per cent by weight of the oxygen given off at the anode has been obtained in the form of ozone. 9 Assuming three volts sufficient to electrolyze the solution, the yield in oxygen per kilowatt hour would be 10 grams, and if 17 per cent of this were ozone, the yield would be only 1.7 grams per kilowatt hour. When com- pared with 70 grams per kilowatt hour, the yield obtained with the silent discharge, these methods are seen to be inefficient from an economical standpoint, though if a high concentration is desired, this can be best obtained by electrolysis. There are two distinct forms of silent discharge of electricity, which differ in their appearance, in the amount of ozone which they produce, and in the current which is required to produce them. 10 If a point one centimeter distant from a plate con- nected to earth is charged negatively to 7000 volts in air, a bluish light surrounding the point can be seen with the naked eye. If the potential is raised, a reddish broad brush appears, separated from the bluish light by a dark space, while the oppo- site plate remains dark. These different parts of the discharge correspond to what is observed in a vacuum tube in which the air is at a pressure of a few millimeters of mercury. The bluish light corresponds to the negative glow, the dark space to the Faraday dark space, and the reddish light to the positive column of light. With a positively charged point and a low potential differ- ence, a reddish envelope of light is first observed, from which a brush is developed on increasing the potential. The ability 9 Fischer and Massenez, Z. f. anorg. Ch. 52, 202, (1907). 10 Askenasy, Technische Elektrochemie, p. 240, (1910); and Warburg, Ber. d, deutsch. phys. Ges., (1904), 209. 292 APPLIED ELECTROCHEMISTRY to form this brush is important for the ozone formation, and is lost by points after use. In place of it a spark discharge is produced ; but the brush discharge can be produced even on old points by placing a spark gap 0.1 millimeter long before the point. If the discharge takes place between paral- lel conducting plates, either one or both being covered with a dielectric, the case is more complicated. 11 This type of ozonizer was devised by W. von Siemens and is usually called by his name. 12 Siemens's original ozon- izer consisted of concentric tubes, as shown in Figures 124 and 125. Two such tubes, with the sides a and d covered with a conduc- tor, such as tin foil, may be looked upon as a series of condensers connected in series, with an ohmic resistance in parallel with one of them. In this case there would be three con- densers : ab, be, and cd ; while if the inner tube is bare metal there would be only two : be and ab. When the space be is filled with a perfect insulator or with a perfect conductor, the current has its small- est or its largest value, re- spectively. In both cases the apparatus is a perfect condenser and absorbs no energy, since cos = 0, FIG. 124. Longi- &J ' . V. ' tudinai section of where $ is the angle of Siemens's original p nase difference between FIG. 125 Transverse sec- the voltage and the cur- *" M .'.o,ir rent. For an average conductivity in 6 between and 1 may occur. With increasing current strength cos $ decreases, probably because the resistance of the gas decreases with increasing current. If the frequency of the alternating current increases, cos in- creases and approaches 1. High frequencies, between 200 and 500 per second, should therefore be used. Ozonizers with one tube bare metal are better than those with both tubes glass, for cos (j) is larger, and a larger current passes, for a given vol- tage, than in a glass apparatus of the same dimensions. If an alternating electromotive force is applied between a point electrode and a plate connected to earth, the positive brush appears. By means of a rotating mirror the positive and negative light can be seen alternately on the point, and its appearance is not much changed when the plate is covered with an insulator. 11 If a Siemens apparatus has a large current passing, a uniform luminosity appears in the space between the elec- trodes, but if the current density is sufficiently lowered, brushes are formed at single points on the electrodes. From the ap- pearance of these discharges, there is no doubt that the same process takes place in the Siemens apparatus as in one with point and plate electrodes, except that in the Siemens ozonizer the effects on positive and on negative points are superimposed, as in the case of a direct current between two metallic points. The production of ozone by the silent discharge of electricity may be considered from the following different points of view : (1) the maximum concentration that can be obtained, (2) the maximum number of grams that can be produced per coulomb of electricity, and (3) the maximum number of grams per unit of power. The latter consideration is, of course, of the most technical importance. As stated above, ozonizers with point electrodes give different results, depending on whether the points are positive or negative to the plate. The Siemens ozon- izer is a third case to be considered. The amount of ozone produced per coulomb is therefore a variable quantity, and fol- lows no known law, such as we have in Faraday's law in the case of electrolysis. In the absence of such a law, it will be necessary to show what the yield is under different conditions 294 APPLIED ELECTROCHEMISTRY and how this is affected by changing the conditions. In ordei to give a systematic survey of this subject, the maximum con- centration will first be discussed for the three cases enumerated above, and the yields per unit of electricity and per unit of power, including the factors that affect them, will then be taken up in the following order, (1) for points negative, (2) for points positive, and (3) for Siemens ozonizers. The Maximum Concentration The silent discharge of electricity has a deozonizing effect on ozone, as well as an ozonizing effect on oxygen. The ozonizing effect of the discharge is proportional to the concentration of the oxygen, and the deozo- nizing effect to that of the ozone. In other words, this reaction follows the mass action law. If the discharge passed for an infinite time, a limiting concentration of ozone would be reached, at which the amount decomposed per second would equal the amount pro- duced. These two different effects have been studied separately by E. Warburg. 1 The experiments were carried out in the apparatus shown in Figure 126. The ozon- izer was connected with an auxiliary ves- sel H by a capillary tube filled with sulphuric acid to a proper distance above B. and H each had a volume of a little over one cubic centimeter. The point electrode e l was a platinum wire 0.05 millimeter in diam- eter; the earth electrode 2 was a platinum wire 0.5 millimeter in diameter bent in the form of a U to increase the surface. After filling the appara- FIG. 126. Apparatus for determining the maximum attaina- ble concentration of ozone i Ann. d. Phys. 9, 781, (1900). THE PRODUCTION OF OZONE 295 tus with oxygen and sealing off at ^, 0, and 5, the rate at which ozone was produced, and the concentration, could be observed by the change in the height of the sulphuric acid in the ma- nometer. Table 40 gives the results obtained with e l connected to the negative pole of an electrostatic machine and e z through a galvanometer and to earth. is a constant proportional to the rate of formation of ozone at a given temperature, and a is a constant proportional to its decomposition. TABLE 40 TEMPERATURE PEE CENT OZONE BY VOLUME /3 = A CONSTANT PRO- PORTIONAL TO KATE OF FORMATION o = A CONSTANT PRO- PORTIONAL TO THE RATE OF DECOMPOSITION + 93 1.23 0.0177 1.42 50 2.22 0.0214 0.939 17 3.53 0.0225 0.616 4.19 0.0219 0.503 -71 5.74 0.0232 0.380 This table shows that the maximum concentration decreases as the temperature rises, and that this is due to the increasing decomposing effect of the discharge, and not to a smaller ozonizing effect. This is evident from the values of a and /3. The spontaneous decomposition of the ozone was negligible. The ozonizer was then replaced by the one shown in Figure 127 with a volume of 7.5 cubic centimeters. The point electrode e consisted of a platinum wire 0.05 millimeter FIG. 127. Ozonizer thick, and the earth electrode e 2 was a half cylindrical platinum plate. In this ozonizer the positive, as well as the negative, point discharge could be obtained. In both cases faint, luminous points were visible in the dark on the thin wire, while the earth electrode remained dark. With a current of 33 microamperes the results in Table 41 were obtained. 296 APPLIED ELECTROCHEMISTRY TABLE 41 Point Electrode Negative TEMP. MAXIMUM CONCENTRA- TION PEE CENT OZONE BY VOL. /3= CONST. PROPORTIONAL TO BATE OP FORMATION a = CONST. PROPORTIONAL TO KATE OF DECOM- POSITION 48 2.41 0.00824 0.332 19 3.38 0.00807 0.231 4.45 0.00929 0.198 Point Electrode Positive 48 0.81 0.00243 0.297 19 1.06 0.00258 0.233 1.42 0.00278 0.198 From these results it is evident (1) that the maximum concentration with the point negative is about three times as great as with the point positive ; (2) that this is due to the greater ozonizing effect of the discharge when the point is negative, since the deozonizing effect is approximately the same in both cases ; and (3) the temperature effect is the same for the positive as for the negative point discharge. For the Siemens type of apparatus the limiting concentration of ozone produced from 96 per cent oxygen diminishes slightly with increasing current, as shown by the following table : 2 TABLE 42 TEMPBRATTTBE AMPEKKS x 10 8 OZONE GRAMS PER CUBIC METER PER CENT BY VOLUME 19 19 1.21 3.00 168 165 8.02 7.90 Apparatus changed 24 24 1.33 2.16 114 110 5.46 5.27 2 Warburg and Leithauser, Ann. der Phys. 28, 31, (1909). THE PRODUCTION OF OZONE 297 FIG. 128. Experimental ozonizer Fig. 129. Experimental ozonizer 298 APPLIED ELECTROCHEMISTRY The limiting concentration is evidently a quantity that varies with the apparatus used. The highest value obtained is 211 grams per cubic meter, or 10.1 per cent by volume. 3 Yield per Coulomb for Negative Point Electrode In order to produce the maximum amount of ozone per coulomb, the deozonizing effect of the electric discharge must be excluded. This may be ac- complished by passing the oxy- gen through the ozonizer so rapidly that the concentration of the ozone produced remains very low compared with the maximum concentration attain- able. In a number of the ex- periments referred to below, the concentration of ozone did not exceed one per cent of the maximum. A number of dif- ferent forms of apparatus with a point electrode were used by Warburg in determining these yields. In the apparatus shown in Figure 128, E, the earth electrode, is a platinum plate ; in Figure 129 E is a platinum cylinder ; and in Figure 130, consisting of a liter bottle, E is concentrated sulphuric acid. Figure 131 shows an ozonizer with a number of point elec- trodes. FIG. 130. Experimental ozoiiizer With the point negative, for a given current strength the yield per coulomb is independent Warburg and Leithauser, Ann. der Phys. 28, 25, (1909). THE PRODUCTION OF OZONE 299 of the voltage, as shown by the results of Table 43, 1 obtained with oxygen 93 per cent pure by volume. In the following tables the current given is for one point only, in case the appa- ratus contained more than one point. TABLE 43 AMPERES x 10 VOLTS GRAMS OZONE PER COULOMB 57 4200 0.0375 57.5 9880 0.0386 57.2 11,700 0.0387 FIG. 131. Experimental ozonizer The yield is also independent of the form of the anode, and decreases slowly with increasing current, as is shown by the following results obtained with different forms of anode : l Warburg, Ann. der Phys. 13, 472, (1904). 300 APPLIED ELECTROCHEMISTRY TABLE 44 APPARATUS IN FIG. 128 APPARATUS IN FIG. 129 Amperes x 10 6 Grams Ozone per Coulomb Amperes x 10 6 Grams Ozone per Coulomb 17.4 0.0484 29.1 0.0431 25.1 0.0459 57.5 0.0386 57.2 0.0375 94.2 0.0370 These results are for the case where negative light appears only on the point. If it appears at other parts of the electrode, the yield may increase with the current. The yield depends further on whether the points have been previously used, being greater for previously used points : 2 TABLE 45 Oxygen, 96 per cent pure, by volume TIME DURING WHICH OZONIZER WAS USED BETWEEN EXPERIMENTS VOLTS AMPERES x 10 6 GRAMS OZONE PER COULOMB _^__ 6080 14.6 0.0408 125 min. 6960 21.9 0.0452 165 min. 6420 17,6 0.0803 75 min. 6300 17.5 0.0873 30 min. 8800 17.5 0.0908 This increase in the yield is accompanied by a change in the character of the light on the electrode. When the final state of the electrode has been reached the yield decreases for increas- ing current to a certain point, as shown by the results in Table 46 of experiments with 98.5 per cent oxygen : 3 2 Warburg, Ann. d. Phys. 17, 6, (1905), * Warburg, Ann. d. Phys. 17, 6, (1905). THE PRODUCTION OF OZONE TABLE 46 301 GRAMS OZONE PER GRAMS OZONE PER VOLTS AMPEBES x 106 COULOMB KILO WATT-HOUR 7230 8.83 0.0950 47.3 8800 17.50 0.0908 37.1 12,500 52.30 0.0485 14.0 If the current is increased to a still higher value, the yield reaches a minimum and then increases with the current. This is shown by the results in Table 47, obtained with new points and with oxygen 96 per cent pure by volume : 4 TABLE 47 VOLTS AMPERES x 10 6 GRAMS OZONE PER COULOMB 6080 14.6 0.0423 21.9 0.0340 9610 52.4 0.0307 12,510 130.7 0.0422 7000 21.9 0.0375 In this case also a marked change in the appearance of the light accompanies the increase in the yield after passing the minimum. After a certain amount of practice, it is even pos- sible to predict from the appearance of the light what the yield will be. 6 In changing the temperature and pressure of the gas, not only the substance which is to be acted upon is altered, but also the agent which brings about the reaction ; for the light changes its character when the physical state of the gas through which the current is passed is altered. This fact complicates the study of this subject. The results in Table 48 with oxygen 98.5 per cent pure by volume show how the yield increases with the pressure : 6 4 Ann. d. Phys. 17, 10, (1905). 6 Ann. d. Phys. 17, 7, (1905). e Ann. d. Phys. 17, 12, (1905). 302 APPLIED ELECTROCHEMISTRY TABLE 48 PRESSURE IN MM. OF MERCURY VOLTS GRAMS OZONE PER COULOMB REMARKS 460 3810 0.0365 Points previously sub- 784 5220 0.0522 jected to long use. Current 1210 6900 0.0903 = 37.4 x 10~ 6 ampere 465 5520 0.0355 Fresh points. Current 780 7410 0.0477 = 17.5 x 10~ 6 ampere 1208 9700 0.0597 Between 780 and 460 millimeters pressure, the yield A p for any pressure p is given by the equation 7 A p = A 7QO [1 - (760 -p) 0.00089]. The temperature of the gas in all of these experiments lay between 17 and 23. Table 49 shows the effect on the yield of changing the temperature : TABLE 49 Oxygen 98.5 per cent pure by Volume. Current 37.4 X 10~ 6 Ampere PRESSURE IN MM. OP MERCURY TEMPERATURE VOLTS GRAMS OZONE PER COULOMB 785 16.5 7530 0.0418 782 79.5 6560 0.0376 789 14.8 7500 0.0430 789 80.1 6440 0.0395 788 14.2 7470 0.0430 780 79.7 6320 0.0387 786 15.4 7350 0.0429 786 79.8 6320 0.0394 This decrease in the yield is largely due to the decrease in the density of the oxygen when the temperature is raised. If the pressure is increased enough to keep the density constant, the yield is very little affected. This is shown in Table 50, obtained with points not previously used : 7 Ann. d. Phys. 28, 21, (1909). THE PRODUCTION OF OZONE TABLE 50 303 PRESSURE IN MM. OP MERCURY TEMPERATURE VOLTS AMPERES x 10 6 GRAMS OZONE PER COULOMB 783 s 12.4: 6440 17.5 0.0403 951 80.0 5970 17.5 0.0413 775 11.8 6350 17.5 0.0458 963 80.0 6180 17.5 0.0480 782 10.4 10,190 52.4 0.0389 979 80.3 8860 52.4 0.0370 792 15.0 9470 52.4 0.0394 980 80.0 8920 52.4 0.0388 It is therefore evident that if the density is constant, the yield is changed only a few per cent between 10 and 80 degrees. The relation between the yield per coulomb and the concen- tration of the ozone produced from 98 per cent oxygen is linear. 8 If the concentration is allowed to reach 12.9 grams per cubic meter, the yield falls to 75 per cent of its value for a concentra- tion of 1.3 to 1.6 grams per cubic meter. The formula ^ = 0.166- 0.00215 c gives the yield per coulomb for different values of the concen- tration c between 1.6 and 12.9 grams per cubic meter, and for a current of 0.0175 x 10~ 3 ampere. The yield per kilowatt hour is given by the equation : B = 71.0 - 1.58 c + 0.00090 c 2 . These results were obtained with spheres, in place of points, 1.5 to 2 millimeters in diameter, melted on a wire 1 millimeter in diameter. The yield for this kind of electrode is much higher than for points, and when used as the positive pole, spheres do not show the aging effect that is observed with points. The presence of water vapor in oxygen reduces the yield nearly proportionally to the pressure of the water vapor. 9 The reduction in the yield for seven millimeters pressure is s Warburg and Leithauser, Ann. d. Phys. 20, 734, (1906). 9 Ann. d. Phys. 20, 751, (1906). 304 APPLIED ELECTROCHEMISTRY about 94 per cent of its value for dry oxygen. There is also a great tendency for the formation of sparks when the gas is moist. When oxygen is mixed with only 7 per cent of nitrogen, the silent discharge produces no oxide of nitrogen, 10 but when air is used oxides of nitrogen are produced. The spark discharge produces only oxides, and these prevent the formation of ozone. 11 For air, the yield per coulomb is independent of the voltage for a constant current, as in the case of oxygen, but it is much smaller than for oxygen. This is shown in Table 51. M TABLE 51 Air. Temperature 20. Six Points. Current for One Point = 21.9 X 10-6 ampere VOLTS DISTANCE BETWEEN POINT AND PLATE IN MILLIMETERS GRAMS OZONE PER COULOMB 8240 12.8 0.0112 8300 12.8 0.0110 3930 3.4 0.0110 7950 13.5 0.0110 For air, the yield first decreases with increasing current and reaches a minimum, after which it increases more rapidly than for oxygen, as shown in Table 52. TABLE 52 AMPERES x 10 6 VOLTS DISTANCE BETWEEN POINT AND PLATE IN MILLIMETERS GRAMS OZONE PER COULOMB 21.9 8300 12.8 0.01100 54.5 0.00935 55.2 7830 7.5 0.00766 163.0 12,200 7.5 0.01880 219.0 12,940 7.5 0.02500 10 Warburg, Ann. d. Phys. 13, 470, (1904). 11 Warburg and Leithauser, Ann. d. Phys. 20, 743, (1906). 12 Warburg, Ann. d. Phys. 17, 25, (1905). THE PRODUCTION OF OZONE 305 The change that takes place in the luminosity when the yield begins to increase is similar to that in the case of oxygen. The effect of the concentration of the ozone produced on the yield in air is approximately the same as in oxygen. 13 A, the yield in grams per coulomb, and B, the yield in grams per kilo- watt hour, are given by the following equations, for values of the concentration c between 2.19 and 9.62 grams per cubic meter : A = 0.0780 - 0.00220 c, B = 42.6 - 1.60 c + 0.0036 c 2 . The effect of moisture is greater for air than for oxygen, 7 millimeters pressure of water vapor reducing the yield to 69. 7 per cent of its value for dry air. 14 The effect of temperature on the yield for negative points in air has not been determined. Yield per Coulomb for Positive Point Electrode The effect of increasing the current on a positive point electrode is quite different from the effect on a negative point. With positive points the yield is smallest for small currents, but increases as soon as the positive brush appears, and, with points not previously used, it finally reaches values exceeding the highest ones obtainable with negative points. This is shown in Table 53. l The yield is very much affected by the character of the positive brush, which depends on a number of circumstances difficult to control. The effects of temperature and pressure on the yield with positive points in oxygen have not been investigated. The relation between the yield in grams per coulomb, A, and the concentration of ozone, . The thick- ness of the dielectric has no effect. 4 The following table gives the yield obtained by Warburg and Leithauser in grams ozone per kilowatt hour. 5 TABLE 58 Air GRAMS PER KILOWATT- DISTANCE BETWEEN PLATES IN MM. FREQUENCY HOUR FOR AMPERES PER SQUARE METER Cone. =4 Cone. = 10 grms. per grins, per ma. m3. 2.26 50 78.4 72.2 0.182 2.26 100 81.4 75.7 0.308 2.26 500 66.0 57.1 1.580 4.66 50 62.4 53.3 0.169 4.66 100 63.0 54.0 0.280 4.66 500 58.0 33.0 1.190 4 See also Ewell, Phys. Rev. 22, 243, (1906). 6 Ann. d. Phys. 28, 36, (1909). 312 APPLIED ELECTROCHEMISTRY It will be noticed that the yield for a Siemens ozonizer is considerably higher than for those having a point and a plate electrode, for which the highest value was 36 grams per kilo- watt at a concentration of 4 grams per cubic meter. 2. THE TECHNICAL PRODUCTION OF OZONE Ozone is produced commercially for the purification of water, for bleaching, and for use as an oxidizing agent in organic chemistry. 1 In water pu- rification, the action of ozone is to oxidize the organic matter and to destroy germs. Siemens and Halske make the ozone apparatus shown in Figure 132. 2 The discharge chamber is between two concentric metal cylinders, between la 'Ozone FIG. 132. The Siemens and Halske ozonizer which 8000 volts alternating are applied. The cylinders are Cooling Water T FIG. 133. The Tindal ozonizer immersed in water for cooling, and the outer one is connected to earth. One of the surfaces from which the discharge takes 1 J. W. Swan, Z. f. Elektroch. 7, 950, (1901). 2 Z. f. Elektroch. 10, 13, (1904); Electrochem. Ind. 2, 67, (1904). THE PRODUCTION OF OZONE 313 T Pef,fugrat03r place is covered with a glass dielectric. Air enters at the top, is partly changed to ozone in passing between the walls of the concentric cylinders, and leaves the apparatus from below. The concentration of the ozone is about 2 grams per cubic meter, which is high enough for all ordinary purposes. The yield varies between 18 and 37 grams per kilowatt hour. 3 The Tindal ozonizer is shown in Figure 133. It is in the form of a box, the inner walls of which are water-cooled electrodes and are con- nected to earth. The other electrodes are metal plates inside the box and insulated from it. Be- tween 40,000 and 50,000 volts are applied to the electrodes. The Abraham-Mar- rnier 2 apparatus is shown in Figure 134. It consists of a num- ber of cylindrical, parallel, hollow electrodes of about a square meter area, covered with glass and mounted in a box. Water cir- culates through the electrodes for cooling, be- tween which about 40,000 volts are ap- plied. The Otto apparatus 4 is FIG. 134. The Abraham-Marmier ozonizer The Otto ozonizer FIG. 135. Longitudinal-vertical FIG. 136. Transverse section vertical section , . ,,,. shown in r ig- ures 135 and 136. It consists of a chamber, K, the metal wall, EV of which forms one electrode. The sheet steel rings, $, Askenasy, Elektrochemie, 1, 246, (1910). * Z. f. Elektroch. 7, 790, (1901). 314 APPLIED ELECTROCHEMISTRY sharpened at M, and mounted on an axle on which they rotate, are the other electrode, E%. There is no solid dielectric. Air passes in the box at B arid comes .out at A. While in the box it is ozonized and thoroughly mixed by the rotating electrode. If an arc were to form between the electrodes, it would be ex- tinguished as the grooves RR in the rotating electrode pass the insulat- ing base of the box, aa. About 25,000 volts are applied to the electrodes, the distance between which may be from 10 to 100 millimeters. Small ozonizers are now made for sterilizing water where it is drawn for use, as shown in Figure 137. 5 The transformer and ozonizer are in a metal case, 0. P and S are respec- tively the primary and the secondary of the transformer. The primary is supplied with 100 to 250 volts, which is transformed to 15,000 volts in the secondary. The ozonizer consists of six or more glass plates, 6r, supported on a grooved bracket at the bottom, and by grooved slips at the sides and top. Three pairs of plates, each plate covered on one side with tin foil, are shown in the sketch. The discharge takes place between two opposite FIG. 137. Small ozonizer con- sheets of tin foil one millimeter apart, without an intervening dielectric. The air enters the space between the plates at the top and sides, and is sucked down through the opening at the bottom of the ozonizer by the action of the water at B. The water carries the ozone to .A, where mixture and sterilization take place. The current in the transformer is of course turned on only when water is drawn. 5 Electrochem. and Met. Ind. 6, 304, (1908> APPENDIX TABLE OF ATOMIC WEIGHTS O = 16.00 (1910) 1 (1910) Aluminum* . . Al 27 1 Helium ... . . He 4 Antimony Sb 120 2 Hydrogen . . H 1 008 . ". A 39.9 . . In 114.8 . . As 74.96 . . I 126.92 . . Ba 137.37 . . Ir 193 1 Bismuth. . . . . . Bi 208 . . Fe 55 85 Boron .... . . B 11 Krypton ... . . Kr 83 Bromine . . . . Br 79 92 Lanthanum . . . . La 139 Cadmium . Cd 112 40 Lead . . Pb 207 10 . . Cs 132 81 . . Li 7.00 Calcium . . Ca 40 09 . . Lu 174 Carbon .... . . C 12 00 Magnesium . . . Mg 24 32 C6rium .... . . Ce 140 25 Manganese . . jAtg . . Mn 54 93 ChlorinB ... Cl 35 46 M^ercury . . * . Hg 200 . . Cr 52.0 Molybdenum . . Mo 96.0 Cobalt .... . . Co 58.97 Neodymium . . . . Nd 144.3 Columbium . . Cb 93.5 . . Ne 20. Copper .... . . Cu 63.57 Nickel. . . . . . Ni 58 68 Dysprosium . . Dy 162 5 . . N 14 01 Erbium *-v . . Er 167 4 . . Os 190 9 Europiu m . Eu 152 . . O 16 00 . . Fl 19.0 Palladium . . . Pd 106.7 Gadolinium . . Gallium . . Gd Ga 157.3 69 9 Phosphorus . . Platinum * . . . P . . Pt 31.0 195 Germanium . . Glucinum . . . Ge . . Gl 72.5 9 1 Potassium . . Praseodymium . . . K . . Pr 39.10 140.6 Gold . . Au 197 2 . . Rd 226.4 1 International Committee on Atomic Weights, J. Am. Chem. Soc. 32, 3, (1910). 315 316 APPENDIX TABLE OF ATOMIC WEIGHTS Continued (1910) (1910) Rhodium. , . . . . Rh 102.9 Thallium . . . . . Tl 204.0 Rubidium . . . . Rb 85.45 Thorium . . . . . Th 232 42 Ruthenium Ru 101.7 Thulium . .... Tm 168 5 . . . . Sm 150.4 Tin . . . . . . . Sn 119.0 Scandium . . . . Sc 44.1 Titanium . . . . Ti 48.1 . . . . Se 79.2 Tungsten . . . . . W 184 .... Si 28 3 Uranium ... U 238 5 Silver . . . Atr 107 88 Vanadium y 51 2 . . . . Na, 23.00 . . . . Xe 130.7 . . . . Sr 87.62 Ytterbium . . . . Yb 172.0 Sulphur . . . . S 32 07 Yttrium . . . . . Y 89 Tantalum . . . . Ta 181 Zinc . Zn 65 37 Tellurium . . . . . Te 127.5 Zirconium . . . . Zr 90.6 Terbium . . . . . Tb 159.2 TABLE OF ELECTROCHEMICAL EQUIVALENTS OF THE MORE IMPORTANT ELEMENTS 1 VALENCE MILLIGRAMS DEPOSITED BY 1 AMPERE IN 1 SECOND GRAMS DEPOSITED BY 1 AMPERE IN 1 HOUR . . . Al 3 0.0935 0.3366 . . . Sb 3 0.4152 1.495 ... As 3 0.2589 0.9319 . . . Ba 2 0.7115 0.1976 . . . Bi 4 0.5387 1.939 . . . Br 1 0.8279 2.981 . . . Cd 2 0.5821 2.095 . . . Ca 2 0.2077 0.7476 . . . Ce 3 0.4843 1.744 . . . Cl 1 0.3673 1322 . . . Cr 2 0.2694 0.9696 f| Cobalt .... ... Co 3 2 0.1795 0.3054 0.6462 1.099 Copper .... . . . Cu 3 1 0.2036 0.6586 0.7331 2371 1 Based on the atomic weights of 1910 and on the value 96,540 for the electro- chemical constant. APPENDIX 317 TABLE OF ELECTROCHEMICAL EQUIVALENTS Continued VALENCE MILLIGRAMS DEPOSITED BY 1 AMPERE IN 1 SECOND GRAMS DEPOSITED BY 1 AMPERE IN 1 HOUR Copper Cu 2 Fluorine Fl 1 Gold Au 1 " " 3 Hydrogen H 1 Iodine I 1 Iron Fe 2 " 3 Lead Pb 2 Lithium Li 1 Magnesium . . . . . Mg 2 Manganese Mn 2 " " 3 Mercury Hg 1 Nickel Ni 2 " 3 Oxygen O 2 Potassium K 1 Silver Ag 1 Sodium Na 1 Tin Sn 4 Titanium Ti 4 Zinc . . Zn 2 0.3293 0.1968 2.043 0.6810 0.01043 1.313 0.2894 0.1929 1.073 0.0725 0.1260 0.2845 0.1897 2.071 0.3039 0.2026 0.08287 0.4051 1.118 0.2382 0.3082 0.1251 0.3386 1.186 0.7086 7.353 2.451 0.03758 4.733 1.042 0.6947 3.863 0.261 0.4534 1.024 0.6827 7.457 1.095 0.7290 0.2984 1.458 4.025 0.8576 1.109 0.4504 1.219 NUMERICAL RELATION BETWEEN VARIOUS UNITS ENGLISH AND METRIC MEASURES NOTE. Values taken from "Tables of Weights and Measures," U. S. Coast and Geodetic Survey, 1890. LENGTH 1 meter = 39.37 inches (legalized ratio for the U. S.) 1 meter = 1.093611 yard 1 meter = 3.280833 feet 1 kilometer = 0.621370 mile 1 inch = 25.40005 millimeters 1 foot = 0.304801 meter 318 APPENDIX 1 yard = 0.914402 meter 1 mile = 1.609347 kilometer MASS 1 kilogram = 2.204622 pounds av. 1 grain = 15.43235639 grains 1 pound =0.4535924277 kilograms 1 ounce av. = 28.34853 grams 1 ounce troy = 31.10348 grams 1 metric ton = 1000 kilograms VOLUME 1 liter = 1.05668 quarts 1 liter = 0.26417 U. S. gallon 1 liter = 33.814 U. S. fluid ounces 1 quart, U. S. = 0.94636 liter 1 gallon, U. S. = 3.78544 liters 1 fluid ounce = 0.029573 liter MECHANICAL EQUIVALENT OF HEAT 1 kilogram-calorie (1 kilogram water raised 1 C. at 15 C.) = 427.3 kilogrammeters (at sea level, latitude 45, g = 980.6 c.g.s.) 1 British thermal unit (1 pound of water raised 1 F. at 59 F.) = 778.8 foot pounds at sea level, latitude 45 1 gram-calorie (1 gm. of water raised 1 C. at 15 C.)= 4.190 x 10 7 ergs 1 joule = 10 7 ergs = 0.2387 gram-calorie ENERGY [Winkelmann, Handbuch der Physik, 1, 79, (1908)] 1 kilowatt = 1000 watts 1 horse power (HP) = 550 foot pounds per second = 746 watts = 0.746 kilowatt 1 kilowatt = 1.34 horse power The metric horse power, called in German Pferdekraft or Pferde- stdrke (PS) = 75 kilogrammeters per second = 736 watts Therefore 1 English horse power = 1.014 metric horse power. APPENDIX 319 LEGAL ELECTRICAL UNITS 1 The legal electrical units in the United States are defined as follows : (1) The unit of resistance is the international ohm, represented by the resistance offered to a steady current by a column of mercury at C. whose mass is 0.4521 gram, of a constant cross section, and whose length is 106.3 centimeters. (2) The unit of current is the international ampere and is the equivalent of the unvarying current, which, when passed through a solution of silver nitrate in water, in accordance with standard specifications, deposits silver at the rate of 0.001118 gram per second. The specifications for the practical application of this definition are the following: In employing the silver voltameter to measure currents of about 1 ampere, the following arrangements shall be adopted : The cathode on which the silver is to be deposited shall take the form of a platinum bowl not less than 10 centimeters in diameter and from 4 to 5 centimeters in depth. The anode shall be a disk or plate of pure silver some 30 square centimeters in area and 2 or 3 millimeters in thickness. This shall be supported horizontally in the liquid near the top of the solution by a silver rod riveted through its center. To prevent the disintegrated silver which is formed on the anode from falling upon the cathode, the anode shall be wrapped around with pure filter paper, secured at the back by suitable folding. The liquid shall consist of a neutral solution of pure silver nitrate, containing about 15 parts by weight of the nitrate to 85 parts of water. The resistance of the voltameter changes somewhat as the current passes. To prevent these changes having too great an effect on the current, some resistance besides that of the voltameter should be inserted in the circuit. The total metallic resistance of the circuit should not be less than 10 ohms. Method of Making a Measurement. The platinum bowl is to be washed consecutively with nitric acid, distilled water, and absolute 1 Bulletin of U. S. Coast and Geodetic Survey, Dec. 27, 1893. 320 APPENDIX alcohol ; it is then to be dried at 160 C., and left to cool in a desic- cator. When thoroughly cool it is to be weighed carefully. It is to be nearly filled with the solution and connected to the rest of the circuit by being placed on a clean insulated copper support to which a binding screw is attached. The anode is then to be immersed in the solution so as to be well covered by it and supported in that position ; the connections to the rest of the circuit are then to be made. Contact is to be made at the key, noting the time. The current is to be allowed to pass for not less than half an hour, and the time of breaking contact observed. The solution is now to be removed from the bowl and the deposit washed with distilled water and left to soak for at least six hours. It is then to be rinsed successively with distilled water and absolute alcohol and dried in a hot-air bath at a temperature of about 160 C. After cooling in a desiccator it is to be weighed again. The gain in mass gives the silver deposited. To find the time average of the current in amperes, this mass, expressed in grams, must be divided by the number of seconds during which the current has passed and by 0.001118. In determining the constant of an instrument by this method the current should be kept as nearly uniform as possible and the readings of the instrument observed at frequent intervals of time. These obser- vations give a curve from which the reading corresponding to the mean current (time average of the current) can be found. The current, as calculated from the voltameter results, corresponds to this reading. The current used in this experiment must be obtained from a bat- tery and not from a dynamo, especially when the instrument to be calibrated is an electrodynamometer. (3) The unit of electromotive force is the international volt, which is the electromotive force that, steadily applied to a conductor whose resistance is one international ohm, will produce a current of an international ampere, and is practically equivalent to ^ of the electromotive force of a Clark cell, at 15 C., when prepared accord- ing to the standard specifications. 1 (4) The unit of quantity is the international coulomb, which is the quantity of electricity transferred by a current of one interna- tional ampere in one second. 1 See Bulletin of U. S. Coast and Geodetic Survey, Dec. 27, 1893. APPENDIX 321 (5) The unit of work is the joule, equal to 10 6 (see under Mech. Equiv. of Heat) ergs, and is practically equivalent to the energy expended in one second by an international ampere in an inter- national ohm. (6) The unit of power is the watt, and is practically equivalent to the work done at the rate of one joule per second. NAME INDEX ABRAHAM and Marmier, 313. Acheson, 210, 217, 219, 220, 221. Adams, 34. Addicks, 47, 48, 53, 54. Adolph, 104, 105. Alexander, 203. Ashcroft, 234. Askenasy, 229. Baekeland, 130. Bancroft, 54. Becker, 234. Behr, 22. Behrend, 17. Bennie, 252. Berthelot, 220. Berzelius, 236. Betts, 64, 236. Bindschedler, 76. Birkeland, 279, 282, 287. Bjerrum, 15. Blount, 208. Bodenstein and Katayama, 275. Bodlander, 15. Boiling, 216. Borchers, 57, 75. Bottger, 19. Bradley, 230, 276. Braemer, 289. Bredig, 11, 266, 267. Brown, 55. Brush, 153. Bugarsky, 15. Bunsen, 6, 143, 228, 236. Burgess, 36, 146. Cantoni, 107. Caro, 266. Castner, 131, 233. Cavendish, 270. Coehn, 73. Cohen, 43. Colby, 259. Collins, 2. Colson, 209, 217, 218. Conrad, 207. Corbin, 125. Cowles, 185, 209, 229. Cowper-Coles, 41. Crocker, 126. Cruickshank, 21. Gumming, 15. Daniell, 143. Davy, E., 202. Davy, H., 228, 233, 236. Despretz, 208, 220. Deville, 228. Dietzel, 57. Dolezalek, 154. Donath and Frenzel, 276. Dony-Henault, 78. Easter brooks, 51. Edison, 173, 182. Edstrom, 265. Elbs, 7, 77, 171. Elmore, 40. Endruweit, 41. Engelhardt, 117, 118, 122, 141. Erdmann, 16. Erlwein, 64, 266, 268. Eschmann, 267. Ewell, 311. Eyde, 281, 282. Faraday, 1. Farup, 4. Faure, 153. Ferranti, 259. Finckh, 271, 272. Fischer, A., 21, 25, 27, 29. Fischer, F., 289, 291. FitzGerald, 192, 200, 209, 211, 219, 220, 224, 262. Fodor, 150. Foerster, 6, 24, 37, 45, 50, 55, 62, 67, 72, 78, 79, 81, 82, 85, 88, 91-95, 97, 98, 101-103, 109, 110, 112, 115, 116, 178, 181, 183, 266. Fraenkel, 267. Frank, 266, 268, 287. Frazier, 216. Fromm, 45, 46. Gall and Montlaur, 123. Garuti and Pompili, 140. 323 324 INDEX Gay-Lussac, 83. Gibbs, Walcott, 21, 22. Gibbs, W. T., 123. Girod, 257. Gladstone, 157, 158. Glaser, 106. Goodwin, H. M., 14, 23. Goodwin, J. H., 237. Gray, 310. Gronwall, 250. Grove, 143. Giinther, 45, 67. Guthrie, 11. Guthe, 1. Gyr, 115, 116. Haanel, 242. Haas, 120, 122. Haber, 36, 38, 55, 74, 206, 225, 271, 273, 278, 285, 287. Haeussermann, 131. Hahn, 11. Hall, 229. Hambuechen, 146. Hamilton, 51. Hansen, 199. Hargreaves and Bird, 126. Hasse, 46, 67. Heim, 158. Heimrod, 2, 3. Hering, 187, 192, 198, 261. Hermite, 117. Herold, 181. Heroult, 230, 247, 249, 256. Herz, 77. Hessberger, 282. Hibbard, 252. Hoepfner, 44, 47. Hofman, 49. Holborn, 15. Holland, 173, 177. Holweg, 274. Horry, 206. Houston, 204. Howies, 276. Huber, 276. Hudson, 203. Hunt, 233. Hutton, 229. Isenburg, 76. Jablochkoff, 150. Jacobs, 227. Jacoby, H., 266. Jacoby, M. H., 39. Jacques, 151. Jahn, S., 288, 290. Jakowkin, 80. Jellinek, 272. Job, 11. Joly, J., 11. Jorre, 82, 101; 103. Jost, 286. Keller, 254. Kellner, 118, 121. Kennelly, 173, 184, 204. Kennicutt, 204. Kershaw, 47, 123, 125, 243. Kiliani, 46, 49. Kistiakowsky, 11. Kjellin, 259, 262. Koenig, 273, 274. Kohlrausch, F., 8, 15. Kohlrausch, W., 158. Kretzschmar, 114. Lalande, 144. Lampen, 217. Langbein, 30, 35. Lebedeff, 229. Le Blanc, 1, 22, 23, 25, 26, 69, 75, 76, 107, 109, 159, 161, 267, 272. Le Chatelier, 191. Leclanche, 143. Lederlin, 125. Ledingham, 6. Leithauser, 273, 292-311. Le Rossignol, 286. Lewes, 202, 206, 207. Lewis, 15, 74. Liebenow, 161, 163. Lindblad, 249. Lombard, 267. Lorenz, 94, 109, 233. Lovejoy, 276. Luckow, 75. Luther, 81. Lyon, 251, 252. McDonald, 126. McDougall, 276. Magnus, 52, 54. Marchese, 43, 44. Marsden, 208. Marx, 289. Massenez, 291. Maynard, 59. Metzger, 153. Meves, 79. Moebius, 59. Moissan, 202, 209, 216, 220, 266. Monkton, 242. Miiller, E., 74, 81, 85, 89, 91-94, 97, 108- 110, 112, 117. Muller, F. C. G., 10. Mylius, 45, 46. INDEX 325 Nahnsen, 46. Nernst, 1, 71, 74, 148, 271. Neumann, 232. Nicholson and Carlisle, 6. Northrup, 261. Niiranen, 272. Obach, 141. Oeschli, 99. Oersted, 228. Oettel, 4, 86, 92, 122. Olsen, 232. Ordway, 146. Osaka, 73. Ostwald, 13, 150. Otto, 314. Patterson, 131. Pauli, 115. Pauling, 283. Petavel, 229. Pinchon, 242. Plannhauser, 40. Plante, 152. Priestley, 270. Pyne, 231. Queneau, 191. Rathenau, 237. Rayleigh, 276. Readman, 227. Reichert, 16. Reinfeld, 41. Richards, H. C., 195. Richards, J. W., 123, 132, 204, 222, 277. Richards, T. W., 1, 15, 22. Richardson, 131. Richarz, 7. Rochling and Rodenhauser, 262. Rostosky, 127. Russ, 310. Salom, 72. Salomon, 16. Sand, 81. Saunders, 41. Scheele, 220. Schmidt, 137. Schneider, 144. SchSnbein, 7. Schonherr, 282. Schoop, 141, 181. Schuckert, 119, 122. Schiitzenberger, 209, 217, 218. Seidel, 6. Senn, 65. Seward and von Kugelgen, 238. Siemens, 243. Siemens Brothers, 141. Siemens and Halske, 44, 312. Smith, 22. Soller, 74. Spitzer, 37. Sproesser, 96. Stadion, 99. Stalhane, 249. Stassano, 243, 254. Steiner, 105. Streintz, 160. Swan, 41. Tafel, 70. Taylor, 225. Thompson, 51, 202, 203, 232, 267. Tindal, 313. Tone, 219. Townsend, 129. Tribe, 157, 158. Tscheltzow, 160. Tucker, 203, 217, 220. Turnbull, 257. Ulke, 51, 54, 55. Uslar, 64. Varley, 146. Volta, 143. Walker, 117, 121. Warburg, 273, 290-311. Watson, 153. Wehrlin, 94. Whiting, J., 133. Whiting, S. E., 173, 184. Wilke, 267. Willson, 202. Wilsmore, 14. Wohler, F., 202, 226, 228. Wohler, P., 237. Wohlwill, 50, 57, 58, 61, 112. Wologdine, 191. Zellner, 96. SUBJECT INDEX Alkali hydrate and chlorine ; diaphragm process, 101-104; bell process, 104; mercury cathode process, 106 ; mer- cury diaphragm process, 107 ; bell cell, 131 ; Castner cell, 135 ; Hargreaves-Bird cell, 127 ; McDonald cell, 126 ; Town- send cell, 129; Whiting cell, 133-136. Aluminum, first isolated, 228 ; reduction of oxide by carbon, 229 ; Hall's process, 229-230; Bradley's process, 230; Heroult's process, 230 ; furnace for electrolytic production, 231 ; yield per horse power day, 231 ; temperature of bath, 231 ; table of production, 232 ; electrolytic production as laboratory experiment, 232. Alundum, 227 ; table of production, 228. Bell alkali chlorine cell, 131. Brass plating, 37. Bright dipping bath, 31. Bromate, electrolytic production of, 115. Bromoform, electrolytic production of, 79. Calcium, first isolated, 236 ; electrolytic production by Rathenau, 237 ; by Seward and von Kiigelgen, 238 ; by P. Wohler, 237. Calcium carbide, discovery, 202 ; heat of formation, 202 ; equilibrium with carbon monoxide, 202 ; chemical properties, 203 ; Willson's original furnace for manufacture of, 204 ; first furnace at Niagara Falls, 205 ; Horry furnace, 206 ; yield per kilowatt day, 207 ; raw materials for, 207 ; table of production, 206. Calcium cyanamide, discovery of produc- tion from carbide, 266 ; pressure of nitrogen in, 267 ; chemical behavior, 268; manufacture, 269. Carbon, different forms distinguished, 220. Carbon bisulphide, electrothermic pro- duction; 225. Carbon electrodes, thermal and elec- trical conductivities of, 198-199 ; porosity, 95-96 ; as anodes in alkali chloride electrolysis, 95-97, 105. Carborundum, first produced, 208 ; dis- covery by Acheson, 210 ; named, 210 ; furnace at Niagara Falls for production of, 212 ; reaction of formation, 212 ; raw materials for, 214 ; yield per kilo- watt hour, 214 ; table of production, 215 ; uses, 215 ; analysis of, 210, 216 ; temperature of formation and of de- composition, 217 ; chemical properties, 217. Carborundum fire sand, 211. Chlorate, production by electrolytic dis- charge of hypochlorite, 85 ; current effi- ciency when so formed, 90 ; electro- lytic production in acid solution, 92 ; in alkaline solution, 84, 93 ; current and energy yields, 98 ; Gall and Mont- laur cell for production of, 123 ; Gibbs cell, 123; Lederlin and Corbin cell, 125. Chlorine, chemical action on hydrate, 80-84 ; electrolytic production, see Alkali Hydrate. Chloroform, electrolytic production of, 79. Complex salts in electroanalysis, 25, 28. Conductivity measurement as method of chemical analysis, 15-16. Conductivity, thermal ; method of de- termining for carbon electrodes, 197 ; table of values for refractories, 191. Copper, refining, object of, 47 ; electro- lytic method, 48 ; composition of anodes, 48, 49 ; composition of cathodes, 49 ; composition of slime, 50 ; com- position of fresh electrolyte, 51 ; of foul electrolyte, 52 ; behavior of im- purities in anodes, 49 ; circulation of electrolyte, 52 ; size of tanks, 52-54 ; multiple and series systems of connec- tions, 53 ; effect of temperature on power required, 54 ; voltage per tank, 54 ; polarization, 54 ; cost, 54. Copper, winning of, 43-45. 327 328 INDEX Copper plating, 35. Coulometers, silver, 2-4 ; copper, 4-6 water, 6-11; silver titration, 11-12. Diaphragms, construction of, 75. Edison storage battery, history and con- struction, 173-177 ; table of different sizes, 177 ; theory of, 178-184 ; nickel electrode, composition of, 178 ; po- tential of, 179 ; efficiency of charging 181 ; iron electrode, potential of, 182 effect of mercury contained, 182 ; chemical changes in battery, 183 ; elec- tromotive force of, 184 ; capacity, 184 ; efficiency, 184. Electric furnace, classification, 186 ; design, 192, 199, 200; heat loss through walls, 187 ; through elec- trodes, 193-198. Electrochemical analysis, by potential measurement, 13-15 ; by conductiv- ity measurement, 15-16 ; by titration, with a galvanometer as indicator, 17- 20; by electrolytic deposition, 20- 29 ; change in potential at cathode during, 24. Electrochemical equivalent, 1. Electrode voltage, 196. Electrolytic bleaching solution, see Hy- pochlorite. Electromotive series, 22. Electroplating, 30-34 ; by contact, 34 ; by dipping, 34. Electrotyping, 39. Faraday's laws, 1. Fluoride, electrolysis of, 113. Foil, metallic, electrolytic production of, 41. Galvanometer as indicator in titrations, 17-20. Galvanoplasty, 39. Gas analysis as a means of determining yield of hypochlorite, 86-87. Gold plating, 38. Gold refining, 61-64; cyanide process for extracting from ore, 63. Graphite, first made artificially, 220; theory of formation, 221 ; furnace for, 222 ; table of production, 224 ; thermal and electrical conductivities of, 198- 199 ; electrodes, see Carbon. Hydrogen, electrolytic production of, 137-141; Schmidt cell, 137; Garuti and Pompili cell, 140 ; Schoop cell, 141 ; Siemens Brothers and Obach cell, 141. Hydrogen electrode, in electroanalysis 20. Hypobromite, 114. Hypochlorite, production by action of chlorine on hydrate, 80-84; by elec- trolysis of alkali chloride solution on smooth platinum electrodes, 84-94; effect of temperature, 88; effect of current density, 89 ; prevention of re- duction by chromate, 89; effect of alkalinity on electrolytic production, 93 ; effect of temperature on production in alkaline solution, 94 ; decomposition point of, 109 ; production with platin- ized anode, 94 ; with carbon anode, 95 ; effect of concentration of chloride solution on yield, 89, 96 ; maximum concentration attainable, 98 ; current and energy yields, 98 ; Hermite cell for electrolytic production of, 117; Kell- ner cell, 118, 121 ; Haas and Oettel cell, 120, 122; Schuckert cell, 119, 122. Hypoiodite, 115; electrolytic discharge of, 116. lodate, 115, 116. lodoform, electrolytic production of, 77. Iron, metallurgy of, 239-242; electro- thermic reduction from ores, 242 ; Stassano's preliminary experiments on, 244; Keller furnace for, 246; He- roult's experiments on, 247-249 ; fur- nace of Gronwall, Lindblad, and Stal- shane, 249 ; furnace at H6roult, Cali- fornia, 251. Lead refining, 64-67. Mercury cathode in electroanalysis, 22. Multiple system of connections in metal refining, 53. Nickel plating, 34. Nickel refining, 55-57; Orford process, 56. Nitrogen, fixation of ; by carbide, 266 ; pressure of, in calcium cyanamide, 267; yield of calcium cyanamide per unit of power, 269, 287 ; by oxidation by elec- tric discharge, discovered, 270 ; ther- mal equilibrium in, 271 ; electrical equilibrium, 274; velocity of oxida- tion, 272 ; yield per unit of power, 276, 287 ; apparatus of Bradley and Love- joy for, 276, 277; of Birkeland and INDEX 329 Eyde, 279-282 ; of Schonherr and Hess- berger, 282-283 ; of H. and G. Pauling, 283-285 ; by direct union with hydro- gen, 285 ; equilibrium, 286, 287. Nitrolime, 270. Overvoltage, in electroanalysis, 24 ; in lead storage battery, 170 ; in reduction, 71 ; in oxidation, 73. Oxidation, electrolytic, 73 ; catalytic effect of anode on, 74 ; of chromium sulphate, 74 ; of attackable anodes, 75. Oxygen, electrolytic production of, see Hydrogen. Ozone, discovery, 288 ; heat of formation, 288 ; free energy of, 289 ; velocity of formation by silent electric discharge, 290 ; yields by different methods of production, 291 ; maximum concen- tration by silent discharge, 294-296 ; yield per coulomb, for negative points, 298-305 ; effect of temperature, 302 ; of pressure, 303 ; of concentration of ozone produced, 303 ; of water vapor, 303; of current strength; yield per coulomb for positive points, 305-308 ; effect of current strength, 305 ; of temperature, 305 ; of concentration of ozone produced, 305 ; of water vapor, 306 ; yield with alternating current, 307 ; yield per kilowatt hour for posi- tive and for negative points, 308 ; theory of formation, 309 ; effect of transparency of glass ofozonizer, 310; ozonizer, of Siemens, 292, 309; of Sie- mens and Halske, 312 ; of Tindal, 313 ; of Abraham and Marmier, 313 ; of Otto, 314. Parabolic mirrors, electrolytic produc- tion of, 41. Perbromate, 115. Perchlorate, chemical formation, 84, electrolytic formation, 99 ; technical cells for, 126. Periodate, 117. Phosphorus, 226. Pinch effect, 261. Potassium, electrolytic production of, see Sodium. Potential, at liquid-liquid junctions, elim- ination of, 15. Potential measurement as method oi analysis, 13. Primary battery, denned, 142 ; Volta's Smee's, Grove's, Bunsen's, chromic acid cell, Leclanche's, 143 ; La- lande's, 144 ; Daniell's, 145 ; dry cells, 146 ; Jacques's cell, 151 ; Jablochkoff 's cell, 150 ; ideal carbon cell, 147 ; free energy of, 148. Quicking bath, 38. deduction, electrolytic; denned, 68; reducing power of cathode measured by its potential, 69 ; pressure of hy- drogen corresponding to different po- tentials, 71 ; catalytic effect of cathode on, 72 ; of chromic sulphate, 72 ; of galena, 72. Secondary battery, defined, 142. Series system of connection in metal refining, 53. Silicon, electrothermic production, 219. Siloxicon, 217-219. Silundum, 216. Silver plating, 38. Silver refining, 57 ; Dietzel process, 57 ; Moebius process, 59. Sodium, production by electrolysis of fused hydrate, 233; Castner cell for, 233 ; Ashcroft process, 234 ; uses of, 236; world's production, 235. Steel, electrothermic refining of, 252 ; Stassano's furnace, 254 ; Keller's, 254; Heroult's, 256; Girod's, 257; Kjellin's, 259; Rochling and Roden- hauser's, 262. Storage battery lead, history and con- struction, 152-157 ; chloride cell, 154 ; Gould cell, 156 ; theory of, 157-172 ; chemical changes in, 157 ; change in density of acid on charge and discharge, 158 ; electromotive force, 160 ; tem- perature coefficient, 161 ; Le Blanc's theory of, 161 ; Liebenow's theory, 163 ; charge and discharge curves, 165-167; capacity, 168; current effi- ciency, 169 ; self-discharge, 169 ; sulphating, 171. Tubes, electrolytic production of, 40. Voltameter, see Coulometer. White lead, electrolytic production of, 77. Wire, electrolytic production of, 40. Zinc, electrolytic winning of, 45 ; spongy, 45, 46; refining, 67. Zinc plating, 33. T HE following pages contain advertisements of Macmillan books on kindred subjects MAX LE BLANC'S Text-Book of Electrochemistry Cloth, 8vo, $2.60 nel Translated from the Fourth German Edition by Willis R. Whitney. Professor Le Blanc's treatment covers electrical conductance, electro- motive force, electrolysis, and polarization. He describes solvents that possess power of causing transformation of energy into equiva- lents. He discusses the theory of " electrolytic dissociation " which prevents the neutralization of positively and negatively charged sub- stances. The book closes with a chapter on storage cells, or the ar- rangements by which electrical energy can be accumulated and stored as chemical energy. "The last word to date on the subject of electro-chemistry." Edu- cation. "A reliable and representative statement of modern theory ... in such form as to make the theory applicable to the needs of the re- search chemist for industrial ends." Electrochemical and Metallur- gical Industry. "The English is above criticism and the presentation is lucid and comprehensible to the last degree. There is nowhere the slightest chance for misunderstanding the writer's ideas. ... As a presenta- tion of the fundamental facts and prevalent theories of electrochemis- try the work is probably without an equal, certainly without a superior. As prices go the book is much cheaper than usual an additional recommendation." Science. FRANK HALL THORP'S Outlines of Industrial Chemistry New Edition, fully revised. Cloth, Svo, $3.75 net This book is designed as a text-book for students who have had a good preliminary training in general chemistry, both inorganic and organic, and who desire to become familiar with the processes em- ployed in the manufacture of the more important commercial products. Valuable features are the references to periodicals and the bibliographi- cal lists after each section. " I feel no hesitation in saying that it is the best book for the purpose intended that it has been my good fortune to examine. It fills a very great need for a compact text-book in technological chemistry, and I am sure its use will be extensive. It reaches the standard of Dr. Thorp's usual excellent work in chemistry." Professor CHARLES BASKERVILLE, University of North Carolina. PUBLISHED BY THE MACMILLAN COMPANY 64-66 Fifth Avenue, New York STANDARD BOOKS ON CHEMISTRY PUBLISHED BY THE MACMILLAN COMPANY ABEGG AND HERZ. Practical Chemistry: An Experimental Introduction to Laboratory Practice and Qualitative Analysis from a Physico chemical Stand- point. By R. ABEGG and W. HERZ, of the University of Breslau. Authorized translation by H. B. Calvert, B.Sc. (Vic.), A.I.C. With Three Tables. 12 + nS pages, I2mo, d., $ 1.50 net ARRHENIUS. Immune-chemistry. The Application of the Principles of Physical Chemistry to the Study of the Biological Antibodies. By SVANTE ARRHENIUS. xi + 309 pages, indexes, I2mo, cL, $1.60 net BAILEY. A Text-Book of Sanitary and Applied Chemistry ; or, The Chem- istry of Water, Air, and Food. By E. H. S. BAILEY, Ph.D., Professor of Chemistry, University of Kansas. 20 + 345 pages, I2mo, d., $ 1.40 net BEEBE AND BUXTON. Outlines of Physiological Chemistry. By S. P. BEEBE, Ph.D., Physiological Chemist to the Huntington Fund for Cancer Research, and B. H. BUXTON, M.D., Professor of Experimental Pathology, Cornell Medical College. I 95 pages, I2mo, d., $1.50 net BEHRENS. A Manual of Microchemical Analysis. By Professor H. BEHRENS, of the Polytechnic School in Delft, Holland. With an Introductory Chapter by Professor JOHN W. JUDD, F.R.S., of the Royal College of Science. With 84 Illustrations Drawn by the Author. ii + 246 pages, ismo, d., $1.50 net (postage 8c.) BLOUNT. Practical Electro-Chemistry. By BERTRAM BLOUNT, F.I.C., F.C.S., Assoc. Inst. C. E., Consulting Chemist to the Crown Agents for the Colonies. Second Impression. 11 + 37 3 pages, cL, $3.25 net COHNHEIM. Chemistry of the Proteids. By Dr. COHNHEIM. Prepared from the Second German Edition by Dr. Gustav Mann, Author of " Physiological Histology." 8 + 600 pages, 8vo, $3.75 net COMEY. A Dictionary of Chemical Solubilities, Inorganic. By ARTHUR MES- SINGER COMEY, Ph.D., formerly Professor of Chemistry, Tufts College. 20 + 515 pages, 8vo, d., $5.00 net FLEISCHER. A System of Volumetric Analysis. By Dr. EMIL FLEISCHER. Translated, with Notes and Additions, from the Second German Edition, by M. M. Pattison Muir, F.R.S.E., Assistant Lecturer on Chemistry, The Owens College, Manchester. 10 + 274 pages, I2mo, il., cL, $2.00 net GATTERMANN. The Practical Methods of Organic Chemistry. By LUDWIG GATTERMANN, Ph.D., Professor in the University of Freiburg. Authorized Translation by William B. Schober, Ph.D., Instructor in Lehigh University. Second American from Fourth German Edition. J 5 + 359 pages, i2mo, il., cL, $1.60 net STANDARD BOOKS ON CHEMISTRY Continued LUPTON. Chemical Arithmetic. With Twelve Hundred Examples. BY SYDNEV LUPTON. F.C.S. 12 + 171 pages, i6mo, d., $1.10 net MENSCHUTKIN. Analytical Chemistry. By N. MENSCHUTKIN, Professor in the University of St. Petersburg. Translated from the Third German Edition, under the Supervision of the Author, by James Locke. 12+512 pages, 8vo, cl., $4.00 net MEYER. History of Chemistry from the Earliest Times to the Present Day. By ERNEST VON MEYER, Ph.D. Translated by George MacGowan, Ph.D. 8w, cL, $4.50 net MILLER. The Calculations of Analytical Chemistry. By EDMUND H. MILLER, Ph.D., Professor of Analytical Chemistry in Columbia University. Third Edition. Revised and Enlarged. 10 + 201 pages, 8vo, cL, $1.50 net MORGAN. Qualitative Analysis as a Laboratory Basis for the Study of Gen- eral Inorganic Chemistry. By WILLIAM CONGER MORGAN, Ph.D. (Yale), Assistant Professor of Chemistry in the University of California, (c.) 14 + 515 pages, 8vo, il., cL, $1.90 net NERNST. Theoretical Chemistry from the Standpoint of Avogadro's Rule and Thermodynamics. By Professor WALTER NERNST, Ph.D., of the University of Gottingen. Revised by the Fourth German Edition. 24 + 771 pages, 8vo, cL, $3.75 net NOYES. Qualitative Chemical Analysis, with Explanatory Notes. By ARTHUR A. NOYES, Ph.D., Professor of Theoretical Chemistry in the Massachusetts Insti- tute of Technology. Third Revised and Enlarged Edition. 89 pages, 8vo, cl., $1.25 net OSTWALD. The Scientific Foundations of Analytical Chemistry Treated in an Elementary Manner. Translated by George MacGowan, Ph.D. 31 + 799 pages, 8vo, cl., $2.00 net Manual of Physico-Chemical Measurements. Translated by James Walker, D.Sc., Ph.D. 12 + 255 pages, il., cl., $2.25 net The Principles of Inorganic Chemistry. Translated with the Author's Sanc- tion by Alexander Findlay, M.A., B.Sc., Ph.D. With 122 Figures in the Text. 27 + 785 pages, 8vo, cl., $6.00 net All by WILHELM OSTWALD, Professor of Chemistry in the University of Leipzig. REYCHLER. Outlines of Physical Chemistry. By A. REYCHLER, Professor of Chemistry in the University of Brussels. Authorized Translation by John McCrae, F'h.D. (Heid.). With 52 Illustrations. Second Edition. 16 + 268 pages, cl., $1.00 net STANDARD BOOKS ON CHEMISTRY Continued HEMPEL. Methods of Gas Analysis. By Dr. WALTHER HEMPEL, Professor of Chemistry in Dresden Technische Hochschule. Translated from the Third German Edition and Considerably Enlarged by L. M. Dennis, Professor of Analytical and Inorganic Chemistry in Cornell University. New Edition. 19 + 490 pages, i2mo, il., d., $2.25 net HERTER. The Common Bacterial Infections of the Digestive Tract, and the Intoxications Arising from Them. By C. A. HERTER, M.D., Professor of Pharmacology and Therapeutics in Columbia University, Consulting Physician to the City Hospital, New York. x + 360 pages, index, 8w, d., $1.50 net; by mail, $1.62 net HILLYER. Laboratory Manual : Experiments to Illustrate the Elementary Prin- ciples of Chemistry. By H. W. HILLYER, Ph.D., Assistant Professor of Organic Chemistry in the University of Wisconsin. 6 + 198 pages, 8w, d., 90 cents net JONES. The Theory of Electrolytic Dissociation and Some of its Applications. 12 + 289 pages, i2mo, d., $1.60 net Elements of Physical Chemistry. 11 +565 pages, 8vo, cL, $4.00 net Elements of Inorganic Chemistry. 13 + 343 pages, 8vo, d., $1.25 net Principles of Inorganic Chemistry. 20 + 521 pages, $3.00 net All by HARRY C. JONES, Professor of Physical Chemistry in the Johns Hopkins University, Baltimore. LASSAR-COHN. A Laboratory Manual of Organic Chemistry. A Compendium of Laboratory Methods for the Use of Chemists, Physicians, and Pharmacists. By Dr. LASSAR-COHN. Translated from the Second German Edition by Alex- ander Smith, B.Sc., Ph.D. 19 + 203 pages, 8vo, d., $2.25 net LEBLANC. The Elements of Electro-Chemistry. By MAX LEBLANC, Professor of Chemistry in the University of Leipzig. Translated by W. R. Whitney, Instructor of Chemistry in the Massachusetts Institute of Technology, Boston. New Edition, revised from the Third German Edition. 10 + 282 pages, I2md, d., $1.50 net LENGFELD. Inorganic Chemical Preparations. By FELIX LENGFELD, formerly Assistant Professor of Inorganic Chemistry in the University of Chicago. 9 + 55 P a S es ) I2mo, d., 60 cents net LEWKOWITSCH. Chemical Technology and Analysis of Oils, Fats, and Waxes. Third Edition. Entirely Rewritten and Enlarged. Eighty-eight Illustrations and Numerous Tables. Two Volumes. 16 + 12 + 1152 pages, 8vo, il., d., $ 12.00 net LEWKOWITSCH. Laboratory Companion to Fats and Oils Industries. By Dr. J. LEWKOWITSCH, M.A., F.I.C., Examiner in Soap Manufacture and in Fats and Oils to the City and Guild of London Institute. 12 + 197 pages, 8w, d., $ i.oo net STANDARD BOOKS ON CHEMISTRY Continued ROLFE. The Polariscope in the Chemical Laboratory. An Introduction to Polarimetry and its Application. By GEORGE WILLIAM ROLFE, A.M., Instruc- tor in Sugar Analysis in the Massachusetts Institute of Technology. 7+320 pages, ismo, il., d., $i.po net ROSCOE AND SCHORLEMMER. A Treatise on Chemistry. By SIR HENRY E. ROSCOE and C. SCHORLEMMER, F.R.S. Vol.1 The Non-Metallic Elements. New Edition. Completely Revised by Sir H. Roscoe, Assisted by Drs. Colman and Harden. 12 + 931 pages, 8vo, d., $5.00 net SCHNABEL. Handbook of Metallurgy. By Dr. CARL SCHNABEL, Konigl. Preuss. Bergrath Professor of Metallurgy. Translated by Henry Louis, M.A., A.R.S.M., F.I.C., etc., Professor of Mining at Armstrong College, Newcastle-upon-Tyne. Second Edition. Vol. I Copper, Lead, Silver, Gold. Illustrated with 715 Figures, (c.) 20 + 1123 pages, 8vo, U., cL, $6.30 net SCHULTZ AND JULIUS. A Systematic Survey of the Organic Colouring Matters. Founded on the German of Drs. G. SCHULTZ and P. JULIUS. Second Edition. Revised Throughout and Greatly Enlarged by Arthur G. Green, F.I.C., F.C.S. 10 + 280 pages, imperial 8vo, d., $7.00 net SHERMAN. Methods of Organic Analysis. By HENRY C. SHERMAN, Ph.D., Adjunct Professor of Analytical Chemistry in Columbia University. 245 pages, 8vo, d., $1.75 net (postage 140.} TALBOT. An Introductory Course of Quantitative Chemical Analysis, with Explanatory Notes and Stoichiometrical Problems. By HENRY P. TALBOT, Ph.D., Professor of Inorganic and Analytical Chemistry in the Massachusetts Institute of Technology. 155 pages, 8vo, d., $1.50 net TALBOT AND BLANCHARD. The Electrolytic Dissociation Theory. With Some of its Applications. An Elementary Treatise for the Use of Students in Chemistry. By Professor H. P. TALBOT and ARTHUR A. BLANCHARD, both of the Massachusetts Institute of Technology. 4 + 84 pages, 8w, cL, $1.25 net THORP. Outlines of Industrial Chemistry. A Text-Book for Students. By FRANK HALL THORP, Ph.D., Assistant Professor of Industrial Chemistry in the Massachusetts Institute of Technology. Second Edition. Revised and En- larged, and Including a Chapter on Metallurgy by Charles D. Demond, S.B., Testing Engineer of the Anaconda Mining Company. 26 + 618 pages, 8vo, il., cl, $3.75 net WALKER. Introduction to Physical Chemistry. By JAMES WALKER, D.Sc., Ph.D., F.R.S., Professor of Chemistry in University College, Dundee. Fourth Edition. 12 + 387 pages, 8vo, d., $3.25 net VOUNG. Fractional Distillation. By SYDNEY YOUNG, D.Sc., F.R.S. With 72 Illustrations. 12 + 284 P a S es ) I2mo, il., d., $2.00 net PUBLISHED BY THE MACMILLAN COMPANY 64-66 Fifth Avenue, New York JAMES WALKER'S Introduction to Physical Chemistry Fourth Edition, 8vo, $3.25 net " This volume by Dr. Walker might well be made the basis of a course of lectures, intended to give students who do not mean to specialize in physical chemistry a general idea of the subject, while the same course might be taken with profit as an introductory one by those who expect to go farther in the subject. The author has been very successful along the lines that he has laid down, and his book can be recom- mended heartily." Journal of Physical Chemistry. TABLE OF CONTENTS Units and Standards of Measurement The Atomic Theory and Atomic Weights Chemical Equations The Simple Gas Laws Specific Heats The Periodic Law Solubility Fusion and Solidi- fication Vaporisation and Condensation The Kinetic Theory and Van Der Waal's Equation The Phase Rule Thermochemical Change Variation of Physical Properties in Homologous Series Relation of Physical Properties to Composition and Constitution The Properties of Dissolved Substances Osmotic Pressure and the Gas Laws for Dilute Solutions Deductions from the Gas Laws for Dilute Solutions Methods of Molecular Weight Determination Molecular Complexity Dimensions of Atoms and Molecules Elec- trolytes and Electrolysis Electrolytic Dissociation Balanced Ac- tions Rate of Chemical Transformation Relative Strength of Acids arid of Bases Equilibrium between Electrolytes Neutrality and Salt Hydrolysis Applications of the Dissociation Theory Electromotive Force Thermodynamical Proof Index. WILLIAM OSTWALD'S The Scientific Foundations of Analytical Chemistry Cloth, I2mo, $2.00 net Translated with the author's sanction by George M'Gowan. TABLE OF CONTENTS PART I THEORY. The Recognition of Different Substances The Sepa- ration of Substances Physical Methods of Separation Chemical Separation The Quantitative Determination of Substances. PART II APPLICATIONS. Introduction The Hydrogen and Hydrox'yl Ions The Metals of the Alkalies The Metals of the Alkaline Earths The Metals of the Iron Group The Metals of the Copper Group The Metals of the Tin Group The Non Metals The Calculation of Analyses. APPENDIX Lecture Experiments. PUBLISHED BY THE MACMILLAN COMPANY 64-66 Fifth Avenue, New York HARRY C. JONES'S The Elements of Physical Chemistry Revised Edition, 8vo, $4.00 net *' This new and enlarged edition of Professor Jones's well- known text-book will be welcomed by teachers and students of physical chemistry. . . . This edition has been exten- sively revised, contains a large amount of new matter, and is a satisfactory realization of the author's aim, as stated in the preface, to bring the book up to date. The volume has been increased in size by 85 pages, but the space devoted to new material is somewhat greater than this, a result made possible by judicious omissions. Subjects newly or more fully treated include J. J. Thomson's electron theory ; radio- activity ; the work of Morse, and of Lord Berkeley and E. G. J. Hartley on the direct measurement of osmotic pres- sure ; colloidal suspensions ; hydrolytic dissociation ; conduc- tivity of fused salts ; dissociation in non-aqueous solvents ; catalysis ; the author's hydrate theory ; researches of the author and co-workers on conductivity and viscosity in mixed solvents." American Chemical Journal. " The brief, but carefully prepared, historical sketches, intro- ducing many of the chapters, as, for instance, the one on ' Chemical Dynamics and Equilibrium,' and the great num- ber of references to original papers form excellent features of the book. The marked enthusiasm with which the author writes as a worker in the field is bound to interest the stu- dent in the subject as a living one, with many vital problems yet to be solved." Journal of the American Chemical Society. PUBLISHED BY THE MACMILLAN COMPANY 64-66 Fifth Avenue, New York LOUIS KAHLENBERG'S Outlines of Chemistry Illustrated, Cloth, 8w, $2.60 net A text-book for college students by the Professor of Chemistry and Director of the Course in Chemistry in the University of Wisconsin. " The first five chapters are mainly devoted to experimental work on hydrogen, oxygen, and chlorine as a foundation of fundamental facts and laws for the sixth chapter in which the atomic and molecular theories are presented. After two chap- ters devoted to ozone, hydrogen peroxide, allotropy, and the halogens, in Chapter IX, acids, bases, and salts, hydrolysis, mass action, and chemical equilibrium are concisely defined on the basis of facts. Chapter XVII includes the elements of thermochemistry ; Chapter XX, classification of the elements and the periodic system, and Chapter XXIV, solutions, elec- trolysis, and electrochemical theories. In this arrangement of the essential parts of chemical theory, and with this mode of treatment, it would seem that the author has kept well within the limits of what the average college student can readily com- prehend and assimilate. As stated in the preface, the student becomes a clear logical thinker and he does not look upon the atomic and molecular theories as something arbitrary, meta- physical, and well-nigh incomprehensible ; it is also mentioned that in principle this is the method of Bunsen and of many other successful teachers of chemistry. Historical connections are kept sufficiently in view by frequent allusions. The cuts are clear and well made and the subject matter well printed. ... A critical examination of this work gives the impression, I think, that the author has accomplished his purpose and that he has given us one of the best books and in some respects the best book that has been prepared on this subject." Journal of the American Chemical Society. PUBLISHED BY THE MACMILLAN COMPANY 64-66 Fifth Avenue, New York THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW AN INITIAL FINE OF 25 CENTS WILL. BE ASSESSED FOR FAILURE TO RETURN THIS BOOK ON THE DATE DUE. THE PENALTY WILL INCREASE TO SO CENTS ON THE FOURTH DAY AND TO $I.OO ON THE SEVENTH DAY OVERDUE. OCT 131934 1C 24W40NI FE8 15 1947 \1& OCT2 ^ LD 21-50m-l, J