The Fixation of Nitrogen. - Industrial & Engineering Chemistry (ACS

Ind. Eng. Chem. , 1917, 9 (3), pp 233–253. DOI: 10.1021/ie50087a011. Publication Date: March 1917. Cite this:Ind. Eng. Chem. 9, 3, 233-253. Note: In...
0 downloads 0 Views 2MB Size
I I a r . , 1917

T H E J O C R S - 1 L 0 F I S D C-S T R I A L -4-VD E ,VGISE E R I S G C H E M I S T R Y

233

ORIGINAL PAPERS THE FIXATION OF NITROGEN’ B y J O H N E. BUCHER

The herein described process for the fixation of nitrogen differs primarily from all those now in commercial use in fixing nitrogen in t h e form of alkali cyanides instead of in the form of oxides of nitrogen, calcium cyanamid. nitrides, or ammonia. I t is further characterized by operating a t very moderate temperatures such as 900 t o 9 j o 3 C. so t h a t i t is not dependent upon cheap electsic pori-er and, it can, because of this moderate temperature, be operated in iron retorts. I t is of the utmost simplicity, uses iron. which is the cheapest metallic catalyzer, and does not require pure materials such as nitrogen but can use air or producer gas just as well. I t requires no special apparatus and can hence be operated a t once TTith 1%-hatcan be found in practically every manufacturing community. I t does not require skilled labor t o operate it. and it is preeminently a method which can be installed quickly in an emergency for the preparation of cyanides. ammonia and nitric acid. These statements are based on a very large amount of chemical and engineering work, most of which was done three t o five years ago, and which has appeared only in the form of patents from time t o time. I have not heretofore published anything regarding any part of this work. LIeanwhile certain circumstances arose which made i t necessary t o abandon my work on nitrogen fixation and abnormal political and industrial conditions have affected our country as well as the rest of the world. I think t h a t all thoughtful people agree t h a t we are in no position t o face any serious crisis which might suddenly arise. For example, we are now practically in the midst of a n alkali cyanide famine which is causing very serious hardships in a number of our industries and our Government officials estimate t h a t we would need 180,000tons2 of nitric acid per year in case of war with any first-class power and t h a t i t is of the utmost importance t o have some ready means of getting this from atmospheric nitrogen so as not t o be dependent on t h e hazardous expedient of importing it by ocean transportation in t h e form of sodium nitrate from South America. The popular idea seems t o be t h a t it is necessary t o have cheap hydro-electric power t o provide such quantities of nitric acid together with a costly plant which would require considerable time for its construction, The data already accumulated in m y work show t h a t electric power is not necessary and t h a t t h e process can be installed in a short time on a n y scale desired and a t comparatively small expense. I had not intended t o publish anything on my nitrogen fixation work for a few years more until I could complete 1 Presented before the 9th Annual Meeting of the American Institute of Chemical Enzineers, New York, January 10 to 12. 1917. * Landis. Tronsocltons of the America18 Electrochemicol Society, 29 (1916). 83

some further important engineering work connected with it. The above considerations, however! led me t o the conclusion t h a t I could not in justice delay action any longer, and hence your kind invitation t o present this paper n-as accepted with the hope of completing the work as opportunity offers. HISTORICAL

The fixation of nitrogen in the form of alkali cyanides is by far the oldest of all such methods and its first period of great commercial activity dates from 1840t o 1847 when it terminated in failure. In January, 1839. there appeared a n abstract of the work of Lewis Thompsoni on ”Improvement in the llanufacture of Prussian Blue” with the statement t h a t he had been awarded the Gold Isis medal of the Society of Arts in the pre\-ious pear €or this x-ork. After speaking of the wasteful process of producing cyanides f r o m animal matter then in use, he says: “Reflecting on these circumstances. it occurred t o me t h a t the atmosphere might be made t o supply, in a very economical manner, the requisite nitrogen, if alloved t o act on a mixture of carbon and potash under favorable circumstances. The experiment proved on trial t o be correct, anti in some measure exceeded my expectation, for the carbonaceous matter employed may be worked over again many times, and is even improved by each operation. I found i t necessary t o use iron, for a reason which will be apparent in the explanation of this process: when iron is not employed, a much higher temperature is required.” He ground two parts potash, or pearl-ash, two parts coke and one part iron turnings into a coarse powder and heated the mixture in a n open crucible in a n open fire t o a full red heat for about half an hour, stirring the mass occasionally. He obtained a n abundant yield of cyanide which he converted into Prussian Blue. I regard Thompson’s remarkable work as the most basic t h a t has ever been done on the fixation of nitrogen by the cyanide process. He discloses clearly the idea of using the nitrogen of the atmosphere and also states with the utmost clearness t h a t iron is “necessary” in the process if it is t o be carried out a t a temperature short of “much higher” t h a n a “full red heat.” Thompson’s article soon led t o very active discussions or investigations throughout the scientific world b y some of the most noted investigators of the time, such as Berzelius,* Erdmann and l I a r ~ h a n d , ~ Fownes and Young,4 L a n g l ~ i s . ~ Rielten,6 Delbriick,’ and Bunsen and Playfair.8 Practically all took the perhaps justifiable x-iew



Mechanics’ Magazine, No. 822, May 1 1 (1839), p . 92; also Dingier’s polylechnisches Journal, N. F. 23 (1839), 281. 2 Berzelius, Jahresberichle. 2 1 (1842), 80. Erdmann and Marchand. J . prahl. Chem., 26 (1842), 412. Fownes and Young, J . p r a k l . Chem., 26 (1842). 407. 6 Langlois, A n n . chim. phys., [3] 1, 117. 6 Rieken, Dingler’s poly J o u r . , 121 (1851). 286; Liebig’s A n n a l c n , 79 (1851). 77. 7 Delbruck, Jahresberichle, 1 (1847). 473. 8 Bunsen and Playfair. J . p r o k l Chem., 42 (1847). 392; Report of the British Association, 1845.



234

T H E J O U R N A L OF INDUSTRIAL A N D ENGINEERING CHEMISTRY

t h a t Thompson had not proved t h e fixation of nitrogen because it might have come from the coke and some of those who did experimental work then prepared charcoal from sugar and heated it with alkali carbonate in a current of nitrogen. They all omitted t h e iron turnings and hence either obtained no cyanide, or only traces, or else had t o heat it t o a very high temperature. They finally concluded, however, t h a t nitrogen could actually be fixed in this manner b u t t h a t the favorable conditions for fixation were not known. Commercial work began in 1840 and in 1843 Newton’ took out the first patent for t h e formation of cyanides from atmospheric nitrogen. Works were finally located a t Newcastle-on-Tyne which regularly produced over one ton of yellow prussiate of potash per 24 hours a t a cost of 1.86 francs per pound, by drawing air down through retorts filled with charcoal containing potassium carbonate. These processes completely ignored Thompson’s recommendation of t h e uses of iron and consequently had t o be operated a t a white heat. This caused poor yields, slow action, much loss of alkali, great expense in constantly renewing the refractory clay retorts which were speedily destroyed by t h e alkali and t h e process resulted in great loss of money and failure in 1847. Since this numerous other attempts have been made t o get cyanides from atmospheric nitrogen by methods which in t h e vast majority of cases did not use iron as a catalyzer. I n 1881 t o 1885 Victor Alder2 of Vienna took out a series of patents in Germany and elsewhere. I n t h e first of these he states t h a t alkali carbonates can be converted into cyanides when heated t o redness with carbon in nitrogen and t h a t the process is essentially favored by t h e presence of metals such as finely divided iron, b u t t h a t the use of iron is not a n indispensable condition as t h e process succeeds completely even without its use. I n a second patent he states t h a t t h e combination takes place copiously only in t h e presence of gases containing carbon (hydrocarbons, carbon monoxide, or a mixture of t h e two) in which metals t h a t are able t o transmit carbon, such as iron, manganese, chromium, nickel, a n d CObalt act extraordinarily favorably as will be shown below. These, together with his other statements made in his patents, constitute such a mixture of t r u t h and falsehood t h a t it is not surprising t h a t failure resulted after six years a t t e m p t t o work t h e patents in Germany. Thompson’s original disclosure of the necessity of iron was finally so completely ignored or forgotten t h a t Castnerp who used alkali metals instead of carbonates, stated t h a t iron is “inert” while Acker,‘ who also used alkali metals, disclaimed iron as a “reactive” metal in t h e cyanide synthesis. I n view of these complete contradictions and t h e complete failure of t h e fixation processes depending o n the formation of cyanides it is hence not surprising t h a t development took place along t h e lines of elec1 Bertelsmann, ”Die Technologie der Cyanverbindungen,” Berlin (1906). p. 85. S l b i d . , pp. 90-91. 8 Castner. U . S. Patent 577,837 (1897). 6 Acker, U S Patent 1,019,002 (1912).

Vol. 9, No. 3

trical fixation methods such as the arc and Cyanamid processes as soon as t h e production of large quantitiw of cheap hydro-electric power was accomplished. EXPERIMENTAL I-MAGNESIUM

NITRIDE PROCESS

The above hasty study of nitrogen fixation t o cyanides did not seem very encouraging and I did not have access t o t h e work of Thompson and Alder a t t h e time. Accordingly some work was done with nitrides with a view t o fixing nitrogen according t o t h e following equations: MgClz = Mg Clz ( I ) 3Mg N2 = MgZNz ( 2 ) MgSN2 6NHdC1 = 3MgC12 8NH3 (3) This process was electrolytic in nature and was designed t o affect improvements in t h e ammoniasoda process. The magnesium chloride would be electrolyzed t o magnesium and the resulting metal would be burned in t h e nitrogen from the towers t o fix nitrogen as magnesium nitride. This later would be heated with ammonium chloride obtained from t h e mother liquors t o produce ammonia and regenerate t h e magnesium chloride for t h e electrolytic stage. This would greatly modify t h e ammonia-soda process by fixing its waste nitrogen, recovery of all the waste sodium chloride, elimination of waste liquors containing calcium chloride and t h e production of chlorine as a new product. It depends, however, on cheap electric power and would depend upon being operated in connection with t h e ammonia-soda works. As I wished t o get a general nitrogen fixation process which was not dependent upon any locality or industry, this work was abandoned.

+ +

11-NITROGEN

+

+

FIXATION WITH ALKALI METALS

The decision was now made t o test t h e chemical principles upon which t h e fixation of nitrogen t o cyanide depended and i t was decided t o use free alkali metal a t first t o study t h e reaction: zNa ZC Nz = zNaCN 46,200 calories. (4) An apparatus shown in Fig. I was accordingly con-

+

+

+



of

6

Pipe

FIO.1

structed of a n ordinary ‘/An. malleable iron pipe about 3 0 in. long, connected at both ends with a reducing coupling and with ‘/*-in. inlet a n d outlet pipes. This 1/2-in. pipe was then placed inside of a larger outer pipe which had a tee and bushings a t one end and a ring of asbestos board a t t h e other. This apparatus gave a means of heating in a combustion furnace while t h e inner t u b e was surrounded by a n atmosphere of hydrogen or nitrogen t o prevent oxides of carbon from diffusing through t h e walls of the inner tube and forming sodium carbonate with the metallic sodium. Hydrogen diffuses into t h e inner tube SO easily t h a t , with a slow current of nitrogen and a 2 - f t .

hfaT., 1917

T H E JOURNAL OF INDUSTRIAL A N D ENGINEERING CHEMISTRY

section red hot, the exit gas a t the end of t h e t u b e burned and contained about zoo cc. of hydrogen per hour. Carbon monoxide diffuses very much more slowly and was detected with iodine pentoxide. EXPERIMENT I-The inner tube was t h e n filled with lampblack containing substantially no ash a n d t h e whole heated in a current of hydrogen t o a white heat, cooled, opened, and then 14 grams of sodium added a t one end and a current of nitrogen passed into t h e inner tube. Reaction began slowly and continued for three days ( z 5 hours), t h e current of gas being occasionally reversed t o drive sodium vapor first one way and then the other. The tube was cooled, opened and found t o contain perhaps 0.5 gram of unchanged metallic sodium. The tube and contents were washed with water a n d gave 23.7 grams of sodium cyanide which is 79 per cent of the theory based on t h e quantity of sodium used. I n another case t h e inner tube was charged with 80 grams of electrode graphite, powdered t o pass through a Ioo-mesh sieve, and ignited with a current of hydrogen for over one hour. Then it was cooled, 1 2 grams of metallic sodium were added and then it was heated in a current of nitrogen for 21/2 days (about 2 0 hours) t o redness as above and 58 per cent of t h e theoretical amount of cyanide was obtained. Quite a number of experiments of this nature were made a n d while t h e absorption of nitrogen was sometimes faster t h a n a t others, it was always a matter of hours. There is hence no question t h a t cyanide formation is slow under these conditions. EXPERIMENT 2-1 now decided t o t r y my idea t h a t iron should act a s a catalyzer notwithstanding t h e above assertions t o t h e contrary. A l/*-in. iron tube was hence heated as before while a current of nitrogen was passing, after being charged with 1 2 0 grams finely powdered alcoholized iron, 12 grams of ignited lampblack and after 7 grams of metallic sodium had been pushed into t h e charge. When t h e tube got t o a low red heat, absorption of nitrogen began so rapidly t h a t a partial vacuum was formed and t h e water through which t h e exit gas bubbles started t o rush back towards t h e hot iron tube a n d was stopped only by quickly turning on t h e nitrogen in a torrent. T h e absorption was practically instantaneous a n d there was no time t o take observations. Only a small quantity of exit gas bubbled through t h e water a n d i t must have been argon. T h e whole thing was finished just as soon as t h e requisite nitrogen could be passed in a n d 94 per cent of t h e sodium was converted into cyanide. The absorption was exceedingly sharp a n d t h e end of t h e t u b e was entirely free from carbon and contained a core of porous iron of a bright silvery luster and so malleable t h a t it flattened out under a pestle. There were sintered globules of metal showing t h a t t h e temperature must have risen hundreds of degrees higher inside of t h e tube t h a n outside owing t o t h e powerful exothermic nature of the reaction, This experiment shows with t h e utmost sharpness t h a t : ( I ) Iron is a n exceedingly active catalytic agent i n t h e fixation of nitrogen.

13 5

(2) The reaction is powerfully exothermic. (3) We have a new method for t h e preparation of argon. (4) The reaction gives a fine method of separation of nitrogen and t h e argon group in gas analysis. We may now write the equation thus:

+

+

+

+

2Na zC Nz Iron = zNaCN Iron (5) Suspecting t h a t much of t h e slow absorption in Experiment I was due t o t h e walls of the iron tube, the experiment was repeated by forcing a thin seamless copper tube into t h e inner iron tube. This tube was charged with 4 grams of sodium and some’ ignited lampblack and heated in a current of nitrogen for 4 hours. No absorption could be noticed and on titration it was found t h a t only I O cc. of nitrogen had been converted into cyanide. This shows very sharply t h e catalytic influence of the walls of iron tubes. 111-PURIFICATION

OF IRON

The above Experiment 2 and the equation shows t h a t while we are dealing with t h e fixation of nitrogen with sodium, carbon and iron, we are necessarily removing carbon from the charge and hence purifying t h e iron, Sodium in the form of solid, liquid or vapor, has frequently been added t o iron and so has nitrogen. Sodium and nitrogen have even been added separately or alternately t o the same mass of iron. But I have found no disclosure wherein any one even hinted a t treating iro3 with sodium and nitrogen simultaneously so as t o remove the carbon in t h e form of cyanide by virtue of the catalytic action of the iron itself. This seems t o be a novel process which is just as interesting as t h e nitrogen fixation itself. It depends upon a powerfully reducing action while all large commercial processes depend upon t h e oxidizing action for the removal of carbon. It cannot attack t h e iron after the carbon is removed as is the case in oxidation reactions. It will remove not only carbon, b u t also sulfur, oxygen and phosphorus. It suggests t h e idea of removing silicon and manganese by the oxidation processes even t o the extent of overburning t h e iron and then removing t h e remaining oxides, carbon, sulfur and phosphorus with sodium with or without t h e previous addition of carbon and t h a t a Bessemer converter might be blown with nitrogen and sodium vapor. It also leads one t o wonder whether with very cheap sodium it could not be made t o apply t o iron from ores very rich in phosphorus. T o further test these ideas for solid iron, a number of hacksaw blades were heated t o redness in nitrogen and sodium vapor for a number of hours. They came out very soft and silvery in appearance and could not be tempered, thus showing t h a t the carbon had disappeared and t h a t t h e iron had acted as a catalyzer t o purify itself. This removal of carbon, sulfur, and phosphorus must take place substantially instantly a t t h e surface of the iron and it is hence determined practically by the rates of the diffusion of these substances through t h e hot iron. The removal of carbon from a piece of steel wire is so complete t h a t when the latter is made the anode in dilute hydrochloric acid, it remains bright and there

T H E J O U R N A L OF I N D G S T R I A L A N D ESGINEERIAVG C H E M I S T R Y

236

is'not t h e slightest sign of carbon as t h e iron dissolves. Ifldesired t h e surface of t h e iron can be purified quickly a n d the interior will remain in t h e form of steel. Many interesting topics arise regarding t h e electrical properties of iron thus purified, corrosion of iron and t h e coating of t h e metal with tin, zinc, etc., but their discussion would require a separate paper. I V-P

R E P A R A T I 0 h- 0 F AI E T A L L I C S OD1 U 31

The general principles of physical chemistry regarding t h e reversibility of chemical reactions a t once led me t o consider whether t h e above reaction for the decarburization of iron might not be rei-ersible and t h a t it might hence be written: 2SaCN

+ I r o n s a S a + N2 + Carburized Iron

(6)

You will a t once notice t h a t this is simply the equation for t h e case-hardening of iron by means of cyanides which has been known for a long time. I n fact, it is practically evident t h a t when t h e iron takes t h e carbon from sodium cyanide, i t must set free t h e other two elements, sodium and nitrogen, which do not recombine under t h e conditions a t which case-hardening takes place. This hence gives us a new method of getting metallic sodium based simply on t h e process of case-hardening iron with t h e additional precaution of keeping away air or gases which would destroy t h e liberated metallic sodium. With all t h e practical use of cyanogen compounds in casehardening. i t is curious t h a t no record of such a disclosure for preparing sodium could be found. EXPERIIIEKT 3-Accordingly some sodium cyanide was mixed with pure iron powder in a n iron tube and heated. A considerable layer of metallic sodium was found in t h e colder part of t h e tube. Similar experiments with potassium cyanide gave metallic potassium quite readily. T h e same thing happens when t h e cyanide is added t o molten iron and this might be of special interest practically because t h e carbon could be removed from t h e molten iron by a current of air blown into t h e converter and cyanide then passed in again as before. The general method of operation would be very similar t o t h a t used in making water gas commercially. I did not follow this up because I had no suitable means for testing t h e engineering features of such a process and t h e method of getting sodium by t h e electrolysis of cyanide with t h e incident formation of cyanogen, oxamide, oxalic acid and formic acid, which will be described later, seems t o offer greater advantages t h a n this new method. V-KITROGEK

FIXATION

W I T H ALKALI CARBONATES

Having now determined t h a t iron, or some similar element, was a n exceedingly efficient as well as essential catalytic agent in t h e fixation of nitrogen, I decided t o test t h e following ideas : A t a given temperature which is quite high, sodium carbonate and carbon give the reaction

+

+

Na2C03 2 C Z z N a 3CO (7) which was formerly used for t h e preparation of sodium. It would hence seem t h a t a t moderate temperatures

T-01. 9, S o . 3

of, say, from 860 t o 980' C., we might get traces of metallic sodium according t o t h e equation. This free sodium should then react practically instantly with t h e carbon and nitrogen, providing finely divided iron were present as has already been shown. If these ideas are correct, we should get a n extraordinarily cheap and efficient method of nitrogen fixation. providing t h e reaction velocity for t h e formation of metallic sodium vapor is high. This would mean t h a t as t h e sodium vapor disappeared t o f o r m sodium cyanide, more sodium would immediately be liberated, according t o t h e law of mass action, t o re-establish t h e equilibrium and t h e process would thereby become continuous, and we could represent it thus: 2NaCS

+ 1C +

=

2SaCS

S?

+

3 C 0 - 138,500 calories

(8)

The following two experiments support these i-iews in t h e most striking manner: E X P E R I ~ I E S T 4-In this experiment a mixture of I O grams of finely poivdered graphite was heated with j grams of sodium carbonate from 920 t o 940' C. for 50 minutes in a current of nitrogen. A '/*-in. copper tube was used in order t o avoid t h e catalysis of iron walls. At this temperature there was a very slow b u t steady evolution of carbon monoxide and some white fumes passed out of t h e tube with t h e gas current. These fumes imparted a steady b u t not very intense yellow color when they were led into a Bunsen flame. Upon cooling, a minute quantity of free metallic sodium was found in t h e end of t h e copper tube. This hence gives positive experimental evidence t h a t t h e formation of metallic sodium from carbonate takes place according t o t h e above equation a t temperatures below 940' C. The contents of t h e tube were lixiviated a n d tested for cyanides with iron salts in t h e usual way. No precipitate of Prussian blue formed even on standing for a few minutes and only a sljght greenish blue color showed no more than a trace of cyanide t o have been present. A similar experiment was then made m-itli 3 0 grams of finely divided iron a n d j grams of sodium carbonate in a slow current of nitrogen for j o minutes a t 920 to 1000' C. In this test t h e issuing gas did not impart t h e slightest color to t h e Bunsen flame, thus showing t h a t no sodium was formed and t h a t t h e carbonate is so slightly volatile as not t o give a flame test under the conditions of the experiment. E X P E R I M E K T j--4 mixture of I O grams of graphite a n d I O grams of finely powdered iron with j grams of sodium carbonate was now heated in a 1/2-in. iron tube from 9 2 0 t o 940' C. for j o minutes in a current of nitrogen, just as in the preceding experiment in which n o powdered iron was used. There was a steady flow of gas which burned with t h e characteristic blue flame of carbon monoxide. The escaping gases showed no fumes nor did they a t any time show t h e slight yellow color. The product showed a conversion of over 60 per cent sodium carbonate into sodium cyanide and a very heavy precipitate of Prussian blue was obtained.

hIar., 1917

T H E JOUR.Y.4L

O F I S D U S T R I A L A N D E S G I N E E RING C H E M I S T R Y

237

These two experiments establish t h e foundation for I O i t . in length. Time will not permit t h e description nitrogen fixation in t h e sharpest possible manner a n d of these experiments a n d tests. I can say t h a t only they support t h e above quotation of Lewis Thompson’s t h e best results were obtained with a-in. pipes, I O ft. work in t h e most striking way. They show t h a t a t in length, I n some of these cases equal portions of t h e a red heat we can have no nitrogen fixation with carbon solution, lixiviated from t h e cyanized charge, required a n d soda ash alone, b u t , t h a t when finely divided iron t h e same number of cc. of N / I O AgN03 solution, acidified with nitric acid, and of N ’ I O H2S04 (with is added, we have a n exceedingly efficient method. These experiments, which cost practically nothing methyl orange) for titration. These figures show all t h e alkali metals t o be present and took only a few hours’ time, were made in 1912 as cyanides, a n d this is what I mean when I speak of a n d they emphasize t h e utter inexcusability of those cyanide of IOO per cent purity. T h e n such a solution who quoted Thompson’s invention and then ignored his positive statement about t h e necessity of the pres- is evaporated t o dryness in “ L ’ U C U O , it is easy t o obtain ence of iron for work a t a red heat. They also show a residue directly which titrates for over 9 j per cent t h a t Alder’s above-mentioned statement, t h a t t h e fixa- of sodium cyanide. BRI Q UETTIKG tion of nitrogen a t a red heat succeeds even without Having thoroughly established t h e general condit h e presence of iron, etc., is entirely erroneous and t h a t his claim t h a t “carbonizing” gases, such as carbon tions for efficient nitrogen fixation with substances monoxide, must be introduced with t h e nitrogen if in powder form, t h e next thing was t o work out engineercyanides are t o be formed abundantly, has not t h e ing methods for handling t h e materials. The inconvenience of heating powder in a current of nitrogen slightest basis in fact. It shows a t once t h a t , b y ignoring t h e work of Thomp- gas a t a red heat, naturally suggests briquetting. son, investigators a n d manufacturers were forced t o ,4 s t u d y of t h e art reveals numerous examples of briwork a t such exceedingly high temperatures t h a t t h e quetting, usually done with such substances as tar, results could be only a ruinous destruction of apparatus pitch, resins, etc., or by use of lumps of charcoal, wood, a n d like substances t o attain t h e desired porosity, a n d commercial failure. Desiring t o avoid t h e use of such foreign substances, T h e above facts are in harmony with Dr. Ewan’s statement in his carefully prepared article in Thorpe’s I first prepared briquettes by heating t h e mass of (I Dictionary of Applied Chemistry,” Revised Edition, sodium carbonate, coke and iron t o t h e melting point of soda ash in t h e absence of air. The pasty powder Volume 11, 1912, page 196: “A few temperature measurements which t h e writer upon slight pressure becomes compacted and yields made with a platinum rhodium thermocouple showed briquettes which are very hard and compact upon t h a t potassium vapor is first evolved from a mixture cooling. I have found such briquettes t o be quite of potassium carbonate and Gharcoal a t about 13 joo active. and they gave cyanide of 100 per cent purity a n d t h a t t h e formation of cyanide takes place very upon being heated for I O minutes in a current of slowly a t this temperature, t h e potassium cyanide nitrogen. volatilizing for t h e most p a r t . ” This is a fascinating process since t h e briquettes These experiments, together with Dr. Ewan’s tem- would already be at a cyanizing temperature just as perature measurement, show why el-en the methods they are formed. It would, however, mean t h e solving which used alkali metals without catalyzers were doomed of unique engineering problems; hence I did not follow t o failure as well as those using alkali carbonate. I t it further a t t h e time. The experience with t h e red-hot briquettes led me is of no use from t h e technical point of view t o work on any method which involves conditions under which t o conclude t h a t perfectly satisfactory briquettes material suitable for apparatus will not s t a n d up. could be made b y adding hot water t o t h e mixed I h a d this sharply in mind when I determined to get charge of coke, soda ash, and iron, a n d making use of a cyanide fixation process which should be effective t h e solubility curves and transition points of t h e phases at say from 860 t o 9 j o o C. so t h a t iron or copper ap- of sodium carbonate as shovm approximately in Fig. 2 . paratus might be used in carrying out t h e process. The addition of t h e iron catalyzer solved this problem, which I regard as t h e most important in m y entire work, so completely, t h a t , judging b y Dr. Ewan’s figure, t h e temperature of cyanide formation is lowered more t h a n 500’ C. I made hundreds of experiments with powdered carbon in t h e form of charcoal, coke, lampblack, etc. and with sodium, potassium, caesium, rubidium, I I and barium compounds in t h e form of carbonates a n d /D A 7 & .bo 29 du 7u 80 30 /bo hydroxides t o test various points under widely varying 7iikiEEA7iiE€ /N “C conditions and t o get quantitative d a t a for technical FIG.2 application. Ultimately a series of tests was made with t h e The curve a t a glance shows t h a t t h e mass should powdered materials in horizontal iron tubes of varying be mixed with t h e water a t as near t h e maximum sizes which ultimately reached 6 in. in diameter a n d temperature of 104.75’ as convenient. We shall

138

T H E JOURNAL OF I N D U S T R I A L A N D ENGINEERING CHEMISTRY

Vol. 9 , No. 3

then have a mixture of monohydrate, Na2CO3.Hz0, perhaps a n hour; then t h e soda ash was added and the coke, and iron moistened with a hot saturated solu- grinding continued for 5 minutes or even less. tion of sodium carbonate. These phases cannot change The iron or iron scale a n d coke, being solids, must on cooling until t h e temperature reaches about 35' C., be mixed with t h e utmost thoroughness, b u t t h e soda a t which t h e crystals would take up all t h e water and ash becomes mobile by addition of water in briquetting we would later have a dry hydrate, N a 2 C 0 3 . ~ o H z 0 , as well as by fusing a t 860' C. in t h e furnace so t h a t as far as the water added would permit. no great care is necessary in mixing i t with the other I n other words, there is here a range of fully 70' C. two constituents. Briquettes thus prepared with through which t h e mass might cool before t h e bri- iron scale gave as high as 28 per cent of NaCN in the quetting operation would be interfered with. These heated charge. Thus it is shown t h a t we may start predictions were realized in a n exceedingly satisfactory with iron oxides instead of metallic iron. I n t h e first manner, and many tons of briquettes were prepared run we may thus produce finely divided iron which is without t h e slightest trouble. At first a n ordinary in very fine condition for subsequent runs. I will hereafter, for convenience, speak of t h e surconcrete mixer without steam jacket was used. The face of carburized iron which is in close contact with hot water and t h e heat of hydration of t h e soda ash brought t h e temperature (work done in t h e summer) t h e ground carbon as t h e "catalytic solution surface." high enough t o give fine briquettes, b u t as t h e tempera- For cyanide formation it is only necessary t o have t u r e was not much above t h e transition point, t h e nitrogen, sodium carbonate, a n d such a catalytic whole mass h a d , a tendency t o set like plaster of Paris solution surface in contact simultaneously. It is, hence, evident t h a t sodium carbonate in too large if t h e operations were delayed. quantity will flood t h e solution surface so t h a t t h e Later a steam-jacketed kneading machine was used nitrogen cannot get a t it readily. This will smother with t h e utmost convenience so t h a t t h e full length t h e reaction so as t o make it sluggish, and it is, thus of t h e monohydrate curve was available before cooling evident t h a t there must be a percentage of soda ash could stop t h e process. The hot, dough-like mass was which will give maximum commercial efficiency. briquetted with a n ordinary small power-driven meat T h e three following experiments were made t o give chopper having a cylindrical worm with a steel disc a rough quantitative idea of this factor: They were of 3'/2 in. diameter and having 37 circular holes of made with 50 lbs. of alcoholized iron and with 5 0 lbs. 1/8 in. diameter. The knife was fixed so t h a t t h e bri- of .pulverized coke ground for 5 hours in t h e ball mill quettes were cut off a t about I in. length. The effi- and t h e mixture was then briquetted with three differciency of these small machines was surprising, e. g., ent proportions of soda ash. the 9-gal. steam-jacketed kneading machine gave EXPERIMENT &In this case t h e iron-coke-soda ash 7 2 0 0 lbs. of briquettes per 24 hours with 15 minutes was briquetted in t h e ratio respectively of 2 : 2 : I t o t h e charge, and it could just as well have been run so as t o give a 2 0 per cent soda ash briquette, Then a t 5 minutes t o t h e charge t o give 2 1 , 0 0 0 lbs. of bri150 grams of these briquettes were heated in a I-in. quettes per 24 hours. The meat chopper, which was iron pipe in a current of nitrogen a t a temperature smaller t h a n t h a t often seen in a city market, easily from 600 t o 1000' C. The reaction took place briskly gave 5,ooo lbs. of briquettes per 24 hours with inand t h e issuing gas burned finely. The briquettes experienced help, and I have been able t o force it t o were cooled, lixiviated and titrated in the manner dethree times this capacity for a short interval Thus scribed above. They were found t o contain 13.7 even a few of these machines not much larger t h a n per cent of actual NaCN corresponding t o 85 per cent toys would give a very large yearly output. Of course, of t h e total alkali metal in t h e solution. For brevity, a large brickmaking machine would give a very large I shall speak of such cases as 85 per cent purity. output commercially. EXPERIMENT 7-Here t h e same materials were briThese briquettes should be dried rapidly so as t o quetted t o make a I : I : I iron-coke-soda ash ratio get t h e m hard a n d free from dust. Hot waste gases and 150 grams were heated as before in t h e I-in. are very satisfactory for such drying. A commercial horizontal iron t u b e a t a temperature from 600 t o bake-oven would dry many tons per day or a dryer 1030' C. for 30 minutes. The gas evolution was can be constructed out of brick and iron very quickly. considerably slower t h a n in Experiment 6. The reIf t h e briquettes cool before they are sufficiently dry sulting briquettes contained 2 0 per cent of actual they will harden, and, on further standing, will fall NaCN of 95 per cent purity. t o powder as one might expect when t h e bulky hydrate, EXPERIMENT 8-In this case t h e iron-coke-soda ash Na2COs.IoH20,forms t o break t h e m u p just as freezing inixture was made into I : I : 2 briquettes which conwater may break a spongy, brittle substance. I have tained 50 per cent of soda ash, and 150 grams were dried briquettes in t h e hottest part of t h e Bunsen heated in t h e same horizontal iron pipe in t h e nitrogen flame in less t h a n two minutes. current for one hour a t 600 t o ~ o g o ' C. The combustible gas evolved sluggishly during t h e entire time, The powder was prepared for this work on a commerand t h e mass finally contained 14 per cent of actual cial scale by grinding iron scale (magnetite, haematite, etc., will do just as well) t o a Ioo-mesh powder in a n cyanide of 80 per cent purity. These three experiments show t h a t t h e 2 0 per cent iron ball mill with manganoid steel balls; then an equal weight of coke previously ground in a similar way t o soda ash briquettes were quite reactive, the 33 per cent Ioo-mesh was added and t h e grinding continued for ones intermediate, and t h e 50 per cent one rather

Mar., 1917

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

sluggish so t h a t they h a d t o be heated longer a n d to a higher temperature in order t o get t h e poorer result. These experiments rather indicate t h a t with either larger or continuous acting furnaces i t should be easy t o get cyanized briquettes t o contain a t least 30 t o 3 5 per cent of actual N a C N a n d this prediction was later verified. T h e experiments also clearly show t h e loss of alkali compounds b y volatilization during t h e run. If t h e briquettes do not completely fill a horizontal pipe, or, if they sag somewhat, as inevitably they must as soon as they become plastic when soda ash melts, then t h e nitrogen will t e n d t o pass through t h e upper channel rather t h a n through t h e briquettes. This is further aggravated by t h e counter-current of carbon monoxide which is formed in t h e reaction. This factor is not very serious in a '/*-in. pipe or even a I-in. pipe, b u t i t must become more serious with larger diameters. The following t w o experiments give conclusive d a t a on this point. EXPERIMENT cj--Twenty-five pounds of t h e same 2 : 2 : I briquettes such as were used in t h e I-in. pipe in Experiment 6 were heated in a horizontal 6-ft. piece of 6-in. iron pipe a n d t h e temperature kept from 1000t o 1080' C. for I hour a n d 2 0 minutes while a current of nitrogen was passing. Upon cooling i t was found t h a t t h e plastic charge had sagged in. in t h e upper p a r t of t h e t u b e t o produce a crescent-shaped channel as shown i n Fig. 3.

FIG2

FIG. 4

__c

T h e entire charge averaged only 3.5 per cent of actual N a C N of very low purity as against 13.7 per cent cyanide of 85 per cent purity under t h e far milder and shorter treatment of t h e same sample of briquettes in t h e I-in. horizontal pipe in Experiment 6. A vertical section at about t h e middle of t h e charge was also tested a n d i t showed 12.5 per cent cyanide in a narrowtoplayer, t h e n 3 per cent in t h e broad intermediate layer, while t h e bottom layer showed only 1.4 per cent NaCN. This is also shown diagrammatically in Fig. 3. EXPERIMENT Io-The result in t h e preceding experiment showed t h a t increasing t h e size of a horizontal pipe t o 6 inches, when t h e current of nitrogen passed into t h e end, completely ruined t h e process commercially. If this theory is correct, then this damage could be avoided entirely by placing t h e pipe vertically.

239

This was done with t h e same 6-ft. pipe of 6 in. diameter, a n d as I had no more of t h e 2 : 2 : I briquettes on hand, we used t h e more sluggish I : I : I briquettes as a jo-lb. charge. T h e pipe was heated from 1000 t o 1 0 9 0 ~ C , for only I hour and I O minutes with a rapid current of nitrogen passing. This gave a massive column of yellow flame which at its maximum arose t o a height of 4 t o 5 ft. above t h e mouth of t h e vertical 6-in. shaft. On cooling, t h e t o p of t h e charge was found t o contain 2 4 per cent of actual N a C N while t h e bottom contained 2 0 per cent, giving a n average for t h e entire charge of 2 2 per cent, a n d showing t h a t over I O lbs. of cyanide must have been produced in this run. These t w o spectacular runs on a semi-commercial scale show a t a glance t h a t m y process can be absolutely ruined by simply placing t h e retort horizontally in t h e furnace instead of vertically. The same thing was also shown with longer and larger pipes in which t h e charges were on a scale of hundreds of pounds of briquettes in each run. EXPERIMENT 11-This run was made in t h e same vertical pipe of 6 in. diameter as h a d been used in t h e preceding Experiment IO. It was made t o test t h e volatility of t h e alkali compounds quantitatively: T h e charge happened t o be 55 lbs. of briquettes consisting of 2 2 lbs. of iron scale, 2 4 lbs. coke a n d 9 Ibs. of soda ash. T h e heating continued for 2 hours at 1000 t o I I O O O c. Upon cooling, t h e t o p layer of briquettes was found t o contain 28 per cent of actual N a C N of 87 per cent purity. Passing down, t h e other sections showed, respectively, 16, 14,1 2 a n d I O per cent of actual N a C N a n d t h e purity of t h e four sections varied from go t o 93 per cent. T h e lower section contained a central cone of moderately firm briquettes whose content was 1 2 per cent of N a C N of I O O per cent purity. This cone was surrounded by a n annular section which h a d fallen t o powder because its content of 7 per cent sodium cyanide was no longer enough to hold t h e particles of iron a n d carbon together. These results are shown diagrammatically in Fig. 4, and they show exactly t h e general results we should expect in a n experiment of this sort where we have heat passing in through t h e walls of t h e t u b e and a rapid gas current sweeping through t h e hot briquettes exposing a large surface for t h e evaporation of alkali compounds. T h e endothermic reactions in this case intensify t h e effect. This experiment a t once suggests two methods of solution, both of which I have carried out successfully in practice. The first would be t o make t h e briquettes so active t h a t they could be cyanized a t lower temperatures, t h u s reducing t h e loss by volatility very greatly a n d b y making other suitable changes. T h e other solution would be a continuously operating furnace, and here great volatility would be a most valuable asset. ACTIVITY O F BRIQUETTES

The two following experiments were made with 2 5 grams of briquettes heated in each case in a current

940

T H E J O U R N A L OF I-NDUSTRIAL A N D ENGINEERING C H E M I S T R Y

of nitrogen in a '/*-in. iron t u b e which was laid horizontally into the heat zone of t h e furnace. EXPERIMENT 12-In this case 2 : 2 : I iron-cokesoda ash briquettes were used. They were heated from 710 t o 920' C. in 13 minutes. There was a n exceedingly rapid evolution of carbon monoxide which died down t o almost nothing in 6 minutes before a temperat u r e of 900' was reached. The product contained 1 5 . 2 per cent of actual N a C N of 92 per cent purity. We may, hence, conclude t h a t 2 : 2 : I briquettes should give over ~j per cent cyanide of over 90 per cent purity in I O minutes a t temperatures below 920' C. EXPERIMEXT 13-In this case t h e alkali was washed out of t h e briquettes a n d soda ash then added directly t o t h e moist filter cake of iron and coke so as t o make I : I : I iron-coke-soda ash briquettes, These briquettes did not give t h e rush of carbon monoxide which was noted in t h e preceding experiment, but there was a steady evolution of gas during most of t h e 28 minutes while t h e t u b e was being heated from 620 t o 920' C. There was still some action when t h e heating was stopped. The resulting briquettes contained 30 per cent of actual N a C N a n d t h e purity was 87 per cent. This very satisfactory experiment shows t h a t we can get over 30 per cent cyanide of 8 7 per cent purity in a batch furnace below 920' C. in less t h a n 30 min. heating. We may, hence, conclude t h a t a n efficient nitrogen fixation takes place below 920' C. I n another experiment briquettes from t h e same lot of 2 : 2 : I iron-coke-soda ash briquettes used in Experiment 1 2 were heated for I O minutes from j 6 0 t o 820' C. a n d gave 11.5 per cent of actual N a C N of 61 per cent purity. This shows t h a t even below 820' C. we get considerable quantities in a few minutes. T h e process can, hence, certainly be carried out a t very reasonable temperatures. This reaction could easily be carried out in copper tubes which do not oxidize easily. I t also shows t h a t a t mines t h e cyanide could be lixiriated a n d t h e moist filter cake could be treated with soda ash in t h e kneading machine, briquetted, a n d used over again.

Vol. 9. NO. 3

got into t h e conveyor a n d hardened so t h a t i t could not be turned. This was rather what had been expected; therefore, t h e experiment was modified by using a t u b e 14 ft. long so t h a t t h e worm conveyor could be placed about 4 ft. below t h e bottom of t h e furnace. Fig. j shows t h e principle of t h e arrange-

-

N-

CONTINUOUS FURNACES

T h e second way of overcoming t h e loss of alkali noted under Experiment 11 is t o have t h e charge moving continuously, preferably by gravity; this would keep t h e final concentration of alkali in t h e briquettes just t h e same as at t h e beginning, because, whatever distilled from t h e heat zone would necessarily have t o return t o i t and, hence, would be of advantage in insuring contact rather t h a n a disadvantage. This t y p e of process mould have great advantages in many other ways, some of which will be pointed out. The process would act very much like a blastfurnace b u t i t would differ in yielding a plastic product instead of molten products. T h e first experiment was carried out with a n 8-in. iron pipe, 8 ft. long, placed vertically in a furnace a n d connected b y means of a tee with a worm conveyor immediately below t h e furnace. Everything worked well for a while until t h e hot plastic briquettes

FIG.5

ment. Here t h e descending, incandescent. plastic briquettes would cool by the ascending current of nitrogen, a n d by t h e air outside. The cooling was made more thorough still by using a current of water a n d a piece of cloth wrapped around t h e pipe below t h e heat zone. I n this way t h e briquettes became cooled and hardened so t h a t they passed through t h e conveyor in about t h e same physical condition as t h a t in which they entered t h e furnace. The hardened briquettes, however, stick in t h e pipe so t h a t hammering may be necessary to move them, a n d a t times when t h e pipe was rough it gave much trouble. This can be helped in various ways. such as making the entire pipe conical, or by making only t h a t part below t h e heat zone conical, by making t h e briquettes larger, by feeding briquettes poor in alkali in t h e annular ring next t o t h e wall of t h e pipe, o r

Mar.. 191;

T H E JOUR,VdL O F I N D U S T R I A L A ATD E

even, pieces of coal or coke might be f e d i n t o the outer annular space while the core of the charge contained the briquettes richer in alkali, or, the cooling zone might be a larger pipe t h a n t h a t used for the heat zone as shown in Fig. 6. A complete solution would no doubt be obtained by making the cooling zone so wide t h a t the hot cyanized briquettes would have t o cool below the solidifying point of the sodium cyanide before they could come in contact with the iron walls. The idea is well shown in Fig. 7 where the hot plastic

GI AVE E RI N G C H E MI S T R Y

241

with a thin partition separating the two compartments. An oil burner flame was directed vertically down one flue and t h e flame then led around baffles into the other flue which contained the 22-ft. pipe. The flame then passed up around the pipe and b y controlling t h e draft a t the top a n effective furnace was obtained. I further led a number of air jets into the main flue in such a way t h a t the air oozed through long, narrow slits into the burning gases. I n this may a reducing combustion could be maintained in the heat zone and the entire length of pipe in the furnace could be heated t o a perfectly uniform temperature when desired. The whole furnace was within a derricklike structure high enough so t h a t burned-out pipes could be hoisted out mechanically without cooling them, and new ones inserted. Considering t h a t these were simply first crude attempts, the furnaces vorked surprisingly well. I believe t h a t they are practical and t h a t a gas producer ~ v o u l ddo the work far more easily, b u t , unfortunately. I did not have one-hence, the oil burner furnaces. AX E L E C T R I C F U R S . 1 C E

FIG.6

FIG.i

briquettes are shown not shaded while the cooler hardened briquettes are shaded. I have never had the hot plastic briquettes stick in a furnace operated according t o my directions; the sticking always took place after they had hardened. =Ilso, solid briquettes never stick in a properly operated furnace. This very simple principle shown diagrammatically in Fig. 7 can, hence, scarcely fail t o remove this difficulty absolutely. There is a slight tendency t o stick at t h e top of the heat zone where the briquettes are just softening, but I have also been able t o obviate this with the utmost ease by observing certain conditions. I n some of the work I made tests in a vertical iron pipe 2 2 ft. long and 8 in. in diameter. This required joo lbs. of briquettes t o charge i t and then, in some tests, 3 , 0 0 0 lbs. were fed in a 24-hr. run. The furnace was constructed on a n improved principle as compared with t h a t shown in Fig. j . The furnace consisted essentially of a double chimney

K h e n the price of yellow prussiate of soda and red prussiate of potash began t o rise and ultimately reach values of $ 7 . 0 0 and $ 2 8 . 0 0 , respectively. for each pound of nitrogen they contained. I decided t o construct two types of electric furnaces which depended on the principles I had in mind for years in case of emergency either at the mines or for these chemicals. It was self-evident t h a t they would work and t h a t they could be installed in a few hours where any sort of commercial current is available. They depend simply upon using iron as a resistor and then keeping away oxygen so t h a t they cannot burn out. The first one was merely a piece of I1,’*-in. galvanized iron pipe such as is used on buildings as a water conductor. Two strips of copper were clamped about the ends for leads and the whole tube set vertically in a chimney of brick which were piled on the floor t o give a n opening 9 inches square. The space bettveen the galvanized pipe and the brick were filled with magnesia asbestos and the furnace was complete. I t was filled with briquettes and the currents of nitrogen and electricity were turned on. The zinc burned off like a flash and in a T-ery few minutes there was a steady uniform heat in the briquettes and a fine flame of carbon monoxide. The transformer gave only 3 j o amperes and I had no means of measuring the voltage which was probably not over 2 or 3 volts. Resulting briquettes contained over 2 2 per cent NaCN. Wishing t o carry out this exceedingly satisfactory experiment quickly on a larger scale and being unable t o get a suitable transformer, I purchased a small I 5 kilowatt welding machine for the emergency. These machines are rated a t 2 or 3 volts for the secondary current and only 50 t o 7 j per cent efficiency because they are light and intended only for intermittent service. The furnace %-as now made from a piece of ordinary

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

242

4-in. iron pipe, 6 ft. in length. This was thrust through an iron d r u m about 4 f t . high and thirty inches in diameter with round holes for t h e iron pipe. T h e leads from t h e transformer were connected t o t h e pipe by

+

11

Y

/

WOODEN SUPPORT

-

RECUVER

FIG. 8 strips of copper which had a cross-section of 3 in. by a/4 in. The drum was then filled with magnesia asbestos a n d t h e whole furnace was complete. Then a n &in. t o 4-in. reducing coupling was provided a t t h e bottom of t h e 4-in. pipe and this in t u r n connected t o a 4-ft. length of 8-in. pipe which was closed with a cap a t t h e bottom. A hole for a I-in. pipe was then drilled through t h e t o p of t h e reducing coupling and at one side there was a stuffing box for t h e a/,-in. iron rod which held u p t h e charge. The other end of this rod was supported by a piece of I-inch iron t u b e which served as nitrogen inlet as well. Fig. 8 shows t h e furnace diagrammatically. A piece of fire-brick was now cut and fastened t o another perforated round piece as shown in Fig. 8a. This was lowered into t h e furnace t o support t h e column of briquettes. T h e current available was not over about g kilowatts a n d this was turned on. After several hours t h e pipe had not become quite red hot b u t t h e machine and t h e copper leads were unduly heated. Accordingly a number of buckets of snow a n d ice were thrown over t h e leads and t h e welding machine. When the cloud of steam cleared away i t was found t h a t the iron tube was now a t a good low red heat and it became hotter gradually. T h e treatment with snow and ice was continued

Vol. 9, No. 3

through all t h e runs as it raised the efficiency by lessening t h e resistance of external circuit. Briquettes were now added and a fine run made, but i t took 3 hours because t h e temperature was not as high as it should have been. Even so, I had no trouble in getting charges which gave 2 5 per cent of actual NaCN consistently. When t h e runs were finished a cap was turned on t h e top of the 4-in. pipe and t h e iron rod supporting t h e charge withdrawn. The entire charge dropped into t h e lower 8-in. pipe. T h e iron rod was now pushed back, the top opened and a new brick support lowered into t h e furnace and 1 5 lbs. of briquettes poured in. The whole change took less t h a n z minutes a n d t h e electric current was not shut off. The top was now closed again without interrupting the currents of electricity or nitrogen, the 8-in. pipe was r'emoved, a new pipe substituted a n d t h e run proceeded a s before. T h e current was kept on this for about 3 days and a considerable number of runs made. The iron drum should be of thin metal t o reduce reduction currents. It was so satisfactory in every way t h a t I considered it unnecessary t o run it any longer. This furnace, hence, avoids both sticking of t h e charge and if t h e drum is kept tight, or, a little nitrogen or reducing gas passed into t h e magnesia asbestos, it will be impossible for t h e pipe t o be destroyed by oxidation. By using a large number of pipes in series it would not even be necessary t o provide a transformer, since we could allow a drop of a few volts for each pipe. If one has a three-phase current of, say, 1 2 5 volts a t 1 2 0 amperes, it is possible t o modify t h e furnace shown in Fig. 8 by merely setting a few fire-clay tiles 2 ft. high by 1 2 in. inside diameter around t h e 6-in. iron pipe which I used in this case. I allowed 3 pieces of l/s-in. iron wire each 3 0 ft. long shaped as shown in Fig. 9 t o hang down into t h e annular space between t h e tile and t h e pipe.

TO Supports

+ I I

4 I

I

I

I

I

-L

+

I

I

I

R

There were 3 pieces of fire-brick at the bottom to keep t h e wires spaced from each other and from the iron pipe. The three segments were then used. with

.

Mar., 1917

T H E J O U R N A L O F I S D U S T R I A L AiVD E N G I N E E R I N G C H E M I S T R Y

243

Y-connection and I was thus enabled t o get 18 H. P. in the furnace instead of 1 2 by a single phase. A slow current' of nitrogen carrying some gasoline vapor, was passed into t h e annular space t o prevent oxidation of t h e thin white-hot wire. This worked finely a n d gave over 30 per cent of NaCN in t h e briquettes used. T h e wire, even with this imperfect construction, lasted 9 hours, while it probably would not have lasted g minutes if air had been admitted. On a large scale this furnace would be very simple since we would use only a single phase t o each furnace a n d also use t h e furnaces in series and we would soon have t o use say I-in. iron rods for the resistors even with fairly high voltages such as 2 5 0 t o 500. This entirely practicable type of furnace would probably be less desirable t h a n t h a t described already where t h e pipe itself is the resistor, but, in small scale work, it can be connected a t once t o any city current without expense of loss of time in waiting t o have speciallowvoltage transformers made. The resistance of iron increases probably 5 or 6-fold in heating from t h e room temperature t o t h e cyanizing temperature, and varies with carbonization, etc. I have not been able t o make strictly accurate observations, but in all of my work I found t h a t t h e following formula gave me sufficiently accurate results so t h a t t h e field rheostat regulation gave me entire control over t h e current. Amperes =

500 ~

vw

L2

where V is the voltage used, W t h e total weight of iron in lbs. and L t h e total length of t h e conductor. I found i t t o hold for l/Ic-in. wire, and for rods l / z in. in diameter as well as for pipes from I t o 4 in. in diameter. These electric furnaces are not what I should intend t o install if I were asked t o put u p a large plant b u t they are what I would p u t u p for a n emergency plant (to which I will again refer) because I have personally worked through every phase of these processes and nothing whatever is left t o chance or guess work. T h e time factor would in such a case be more important t h a n t h e greater economy and speed which I feel can almost certainly be obtained by internal electrical heating where the briquettes themselves are t h e resist or, While we are speaking of t h e application of electrical heating with solid catalytic solution surfaces, i t may also be well t o mention t h e experiment with molten iron. Fig. I O shows t h e apparatus a n d is almost selfexplanatory. It represents a cylindrical furnace with basic lining and a perforated bottom like a Bessemer converter. It contains molten iron into which coke or graphite fragments are deeply pressed. The graphite column acts as electric resistor t o produce t h e heat and, while in t h e bath, i t serves both t o ret a r d t h e gas bubbles so as t o insure sufficient contact and t o keep t h e iron saturated with dissolved carbon,

FIG.10

Sodium vapor and nitrogen are blown into t h e bottom of t h e furnace and cyanide distils out a t t h e t o p with t h e argon. The linings I used in t h e short time of t h e run stood better t h a n I had expected. If t h e lining will stand sufficiently in actual work then I feel t h a t ash in the carbon will be the other serious factor since t h e method of getting ashless carbon mentioned hereafter will not apply. However, there are other methods of solution. RATE OF HEAT PENETRATION

I n considering t h e technical application of these inventions, t h e rate of heat penetration into t h e briquettes becomes very important. The following experiment enables us t o form some idea of t h e magnitude of this factor. EXPERIMENT 14-In this run a 4-in. vertical pipe of iron was used with a pyrometer inserted in t h e center of t h e 8-lb. charge of t h e 2 : 2 : I iron-coke-soda ash briquettes through a l/Z-in. iron tube. The whole was placed in a cold furnace and a powerful oil flame used in the heating. By t h e time t h e outer pyrometer read 1030' C. t h e inner one in t h e charge read only 150' C. After 40 minutes more t h e outer one reached 1080' C. while t h e inner one registered only 880' C. I n another 2 0 minutes t h e inside temperature read IOIO' C. while the outer remained constant a t 1080' C .

244

T H E J O U R N A L O F I N D U S T R I A L A N D ELVGILVEERIXGC H E M I S T R Y

The heat, hence, flows in slowly a n d t h e rise is influenced by many factors of which t h e heat absorption of 138,joo calories is one. The result was 17 per cent of N a C N of 99 per cent purity in t h e briquettes. This experiment gives t h e valuable d a t a which enables us t o say t h a t i t would take quite a while t o heat a 6 or %in. column of briquettes t o a uniform temperature. I t also seems t o confirm t h e very probable idea t h a t t h e charge will conduct much better after it is partially heated t h a n when it is cold. The importance of both these observations will be shown later. Ii i T E R 1-AL E L E C T R I C A L HE AT1 X G

Experiment 14 shows t h a t this factor would seriously retard rapid working in cyanide formation with large masses where t h e heat had t o penetrate in from t h e outside. I t would be a question of hours in large tubes. If we could use t h e charge directly as a resistor a n d t h u s generate t h e heat within t h e briquettes themselves, t h e heating would be reduced from hours t o minutes with proportional saving of radiation losses and increase of output. Experiments along this line are very hopeful and wonderfully interesting, and I think no similar thing has ever arisen in electrical furnace work. To discuss adequately t h e very important results already obtained would require more time t h a n is available for this entire paper, and I a m not yet a t liberty t o disclose t h e most important of them.

I a m , hence, compelled t o dismiss this with t h e following general statements: The briquettes contain a n insulator, soda ash, which makes them practically non-conductors a t t h e room temperature b u t t h e conductivity increases very rapidly until a t cyanizing temperatures i t may easily increase 3,000-fold. I t is, hence, only necessary t o heat the charge t o s t a r t t h e current or t o p u t t h e briquettes into a hot furnace. I have heated a charge in a glass tube with the current in a few seconds so t h a t t h e entire run was finished in less t h a n I minute with abundant yield of cyanide a n d I have also heated charges of many lbs. in a few minutes. This suggests wonderful possibilities and shows t h a t here is a n entirely different proposition from t h a t discussed in t h e case of t h e electric furnace shown in Fig. 8.

-

Vol. 9,

KO.3

A number of runs were, hence, made in horizontal iron pipes of I in. diameter as shown in Fig. 1 1 .

Air

I

i

,

o

1Scrleen

!

;

;,I.,,,

/

Briquettes

GoKe

FIG.11

Several inches of layers of coke were held in position in t h e tube with iron screens a n d then another screen kept t h e briquettes away from t h e porous absorbing surface of t h e coke. The pipe mas heated t o t h e cyanizing temperature and a current of air then passed in. Of course, this gave a n extra fine carbon monoxide flame. When t h e process was over the air was shut off before t h e tube was cooled. Similar runs were made in t h e apparatus by using pure nitrogen and the results were just t h e same as when air was used. I n other cases t h e producer gas was made in a separate furnace and then passed into t h e cyanizing tube. There was no apparent difference in any of these results. It is, hence, evident t h a t the process is independent of pure nitrogen and t h a t air, producer gas, flue gas, or even t h e gases from t h e combustion chamber of t h e furnace could be drawn through the cyanizing tube just as well. T o make t h e test still more striking on a commercial scale, two pipes of iron 8 f t . long were p u t side b y side vertically in a furnace as shown diagrammatically in Fig. I 2 .

AIR INSTEAD OF N I T R O G E N

Having now shown how t o fix nitrogen with simple apparatus with pure nitrogen, I desired t o get away from t h e use of pure nitrogen so t h a t my process would not be tied u p t o ammonia-soda or similar works, where nitrogen is a waste product. Also I did not wish t o be compelled t o put up a liquid air or other plant in order t o get nitrogen. I suspected from t h e fact t h a t carbon monoxide is formed in large quantity in t h e process, t h a t a little more would do no h a r m under t h e right conditions. This suggested a t once t h e use of producer gas in place of pure nitrogen or even t h e use of air directly providing i t passed through hot coke before it reached t h e catalytic solution surf ace.

FIG. 12

One pipe was 6 in. in diameter and as I had no other of this size, t h e other was taken of 4 in. diameter. The

Mar., 1917

T H E JOLTRIVAL O F I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y

pipes were closed with caps a t t h e bottom a n d t h e n iron pipes of '/2 in. diameter were thrust t o t h e bottom of t h e larger pipes a n d about a foot layer of coke was dropped into each one. Then 7 2 lbs. of briquettes were poured into t h e larger pipe and an equally deep layer was placed in t h e 4-in. pipe. T h e whole apparatus was now heated red hot for a considerable time a n d pure nitrogen passed into one a n d air into t h e other. The result was 19 per cent of N a C N in t h e pipe which had t h e air and 18 per cent in t h a t through which t h e pure nitrogen passed. These results are practically identical, t h u s showing t h a t air can be used a n d t h a t t h e process can now be operated anywhere. You can now see how easily producer gas could replace t h e nitrogen in working t h e electric furnace shown above in Fig. 8. PRODUCTION OF NITROGEN

Although I do not need nitrogen in m y fixation process with carbonates, i t is, nevertheless, very convenient t o have i t for tests and when working with t h e alkali metals themselves. The passing of air over heated copper has many great advantages, particularly in t h e fact t h a t t h e reagent is solid. Notwithstanding this, there seems t o be much prejudice against it commercially and i t was condemned as impractical or impossible just as many of t h e other basic things I have described to-night.

As no one cited a n y physical or chemical principle against i t which seemed reasonable. i t was in order either t o show t h a t i t could be done or else why i t could not be done. Accordingly a 2-in. iron pipe 40 in. long was filled with t h e turnings which result as a waste product when t h e copper rolls used in t h e textile industry are turned down in a lathe. These turnings can be gathered in quantity a n d at a price below t h a t of copper ingots. The copper used m a y contain considerable zinc and still be entirely suitable for t h e work. Four lbs. were tamped into t h e iron pipe t o give a 28-in. column. These pipes were heated in a combustion furnace t o a low red heat a n d they easily gave zoo liters of nitrogen before t h e y were exhausted. T h e n they were reduced with hydrogen a n d were again ready. I never used gasometers, b u t simply passed t h e air blast furnished by a small glass suction p u m p over t h e hot copper a n d then used t h e gas directly in t h e runs. The arrangement is shown in Fig. 1 2 and shows how easily one can get t h e current of nitrogen automatically a t constant pressure in a n y quantity varying from a few bubbles t o 2 liters per minute. T h e tubes were run for months a n d were just as good as when they were first made. I t was then desired to construct a large commercial unit a n d in order t o test how much sagging, analogous t o t h a t shown by t h e briquettes in Fig. 6, would t a k e place, t h e horizontal pipe was, hence, purposely t a k e n 1 2 in. in diameter b y I O ft. long and charged with t h e

245

copper turnings. It gave 300 liters of nitrogen per minute a t first b u t in a few days was completely useless. Examination showed t h a t t h e copper had sagged 4 in. a n d was, moreover, covered with iron scale from t h e pipe so t h a t t h e air did not get in contact a t all. I n another case, about a 6-in. t u b e was placed vertically b u t t h e plastic copper soon compacted b y its own weight so as t o shut off t h e gas current. These factors having been determined t o be large, it was evident t h a t they must be avoided. This was very easily done and t h e first furnace constructed did SO well t h a t I should simply duplicate it if more nitrogen were needed. I used a vertical pipe I O in. in diameter a n d 6 ft. long, which was heated in a furnace having two vertical flues, one for t h e burner flame and t h e other for t h e tube containing t h e copper. The supports for t h e copper were made from two ordinary I-in. iron crosses with side openings which were connected b y a threaded 6-in. piece of iron rod through t h e side openings. Then 4-in. iron nipples were turned into t h e 8 remaining openings so as t o make t h e 8-armed support which would just slide down into t h e Io-in. pipe. One was dropped in and the copper tamped down just level with t h e t o p of t h e cross with a stick. Then another support was dropped down so t h a t i t rested on t o p of t h e lower cross a n d copper was tamped in as before. This was continued until t h e t o p was reached a n d t h e pipe contained 2 8 0 Ibs. of copper turnings. Every section of copper was thus supported firmly a n d independently. It could not sag from t h e sides because t h e t u b e was vertical. I t could not compact itself because each section had to sustain only its own weight and whatever sag there was in a section could simply open a horizontal section which would do no harm. This furnace was run hard for 6 weeks in t h e tests and I could not detect t h e slightest deterioration. I t gave 16,000 liters of nitrogen in 2 t o 3 hours before requiring reduction a n d then i t was reduced by cautiously adding a number of liters of wood alcohol a t t h e top. This served not only for reduction b u t also as a measure of t h e capacity of t h e furnace. I never heated i t above 4 j 0 ° C. I n actual practice one would reduce t h e copper with t h e carbon monoxide from t h e cyanizing furnace a n d on a large scale perhaps i t might not be necessary t o apply external heat since t h e reactions in t h e copper turnings are exothermic. The reductions should, hence, not be too energetic as otherwise i t might heat t h e copper t o o high. I do not know how others construct their furnaces b u t this one is eminently satisfactory. The expense of copper need hardly be considered because t h e cyanide produced would now pay for t h e copper every 40 minutes and even in normal times in less t h a n half a day. If anyone is doing electrolytic work a n d has waste hydrogen, this will also give nitrogen with t h e utmost ease since i t is only necessary t o burn t h e hydrogen and air in regulated streams in a: very simple apparatus. It consists simply of

T H E J O U R N A L OF I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y iron tubes which may be lined with fire-brick inside in t h e combustion space. This is about t h e same as t h e above scheme with copper, only t h e copper is left out and t h e hydrogen a n d air act simultaneously. The steam formed condenses out and leaves t h e nitrogen which may contain a slight excess of hydrogen if desired and t h e heat generated can be used for evaporation purposes. I A still simpler device follows from my method of changing cyanide into urea as shown below. I n fact, it will be seen t h a t t h e nitrogen there becomes a waste product in nitrogen fixation and it may accumulate twice as fast as it is being fixed. DISTILLATION OB CYANIDE

Having now shown methods of obtaining alkali cyanides, i t is naturally in order t o consider methods of getting t h e cyanide from t h e briquettes. One would wish t o avoid lixiviation if possible and, hence, t h e following work on distillation. As we would destroy t h e catalytic solution surface by heating t h e ironcarbon mixture above t h e eutectic point (about I I 20'C.) because of t h e iron melting t o globules, it seemed worth while t o t r y t h e distillation in either a current of nitrogen or in a vacuum. Both seem t o work. For example: Some cyanized briquettes were placed in a copper tube, heated t o 1000' C. and a t about 2 mm. pressure. The cyanide collected in a pool which solidified t o a clear mass which could not be told from glass by appearance. There was no cyanide whatever left in t h e iron-coke mixture and only t h e merest trace of alkali. I n another case t h e specimen of distilled cyanide was titrated and found t o be 99.9 per cent NaCN. I have also distilled both sodium and potassium cyanides in vacuo while a melting point tube filled with sodium chloride was in t h e mixture. The sodium chloride did not melt; hence the cyanide must have distilled below this melting point (792 or 8 2 0 ' C.). KsFe(CN)a a n d K4Fe(CN)6 were also heated in uacuo. This gave a fine distillate of K C N while iron-carbon in very finely divided form remained. This would be slightly wasteful b u t would work finely since t h e iron, carbon and nitrogen obtained in decomposition would go back into t h e cyanizing process again. ASHLESS CARBON

The great promise of t h e above method of distilling cyanides would soon come t o grief by having t h e ash from the coke accumulate with each repetition until it would stop t h e process. If we could get this carbon from carbon monoxide, it would be ideal. This suggests t h e following equation: 2 C O z C

+ COZ + 38080 calories

The equilibrium for this equation gives almost pure carbon monoxide a t 1050' C. while a t 500' C. it is almost reversed so t h a t carbon monoxide must be unstable a t low temperatures and if we could find a suitable catalyzer we should be able t o get carbon from it.

Vol. 9, No. 3

I first passed pure carbon monoxide over ground coke b u t could notice no change. N o carbon dioxide could be detected with t h e Hempel burette and only a trace of precipitate was obtained with lime-water. Evidently t h e catalysis was negligible. The experiment was repeated as before in a combustion tube of hard glass which had a layer of finely divided iron in t h e bottom, A very striking change took place a t once. The surface of t h e iron blackened a t once and the issuing gas contained over 40 per cent of carbon dioxide in spite of t h e rapid current and t h e very slight contact. The catalysis must, hence, be exceedingly efficient and in a few hours t h e bulk of t h e whole charge had so increased from t h e deposited carbon t h a t the entire section of t h e tube became filled so t h a t t h e gas could no longer pass through. The wonderful importance of this when applied t o t h e iron-coke-soda ash can be seen a t once. It gives us ashless carbon, in t h e form of very finely divided lampblack, together with 38,080 calories of heat developed within the charge itself. I n addition, since t h e coke does not appreciably catalyze t h e reaction, t h e deposited carbon must, hence, be in very close contact with t h e iron. We thus have this automatic process t o do away with t h e ash in t h e carbon, with grinding t h e carbon, with mixing t h e carbon and iron and it provides internal heat in t h e briquettes themselves. It is, hence, equivalent t o t h a t much internal electric heating.

I will show how this will apply later under t h e theory of the continuous furnace. Suspecting t h a t Mondl. might have done such work in connection with his metal carbonyl compounds, I looked up t h e literature and found t h a t he had done so. He found t h a t with 1 5 parts of nickel, he could deposit 85 parts of carbon. I also noticed t h a t I could not obtain over about 43 t o 45 per cent of carbon dioxide in t h e experiments and this suggests t h a t there must be a n equilibrium which should not be here if t h e process were one of pure catalysis since the catalyst should simply hasten t h e velocity of transformation but should not completely change its course. A search of t h e literature revealed considerable work along this line from t h e point of view of the phase rule. Findlay2 discusses the work of Baur and Glaessner, Boudouard and Hahn. Fig. 13 shows a diagram embodying part of their important work on t h e equilibrium relation of carbon dioxide, carbon monoxide, iron and carbon a t varying temperatures, while t h e dotted line in t h e curve shows what the relation should be if no catalyzer were present or if t h e catalyzer exerted no foreign function. It is evident a t a glance t h a t as soon as the concentration of COz reaches a certain value, i t will begin t o oxidize t h e iron t o FeO, thus destroying the catalyzer and completely changing t h e action. The curve A B C of Fig. 13 marks t h e concentrations 1 3

Jour. Am. Chcm. Soc.. 6'7 (18901, 749. Findlay, "The Phase Rule, 3rd Impression," page 306.

Mar., 1917

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

of CO and COZ which correspond t o each temperature as long as any metallic iron is present. T h e maximum value we can get for t h e formation of C 0 2 is hence about 4 2 per cent a t t h e temperature of

680’ C. On t h e other hand, nickel does not form oxides as readily as iron and i t will, hence, transform CO almost quantitatively into COz. This can be shown beautifully as a lecture experiment by placing finely divided nickel in a glass tube heated t o perhaps 300 t o 400’ C. When pure CO is passed into the tube it is so completely transformed t h a t t h e CO flame goes out and t h e exit gas contains 98 per cent of C o t . These phase rule studies are of great importance in t h e study of blast-furnace reactions, case-hardening, etc., and I will further apply them under t h e question of t h e oxidation of iron pipes. O X I D A T I O N OB I R O N P I P E S

I n conducting t h e above process by external heating, t h e question of oxidation of iron pipes must be seriously considered, but, from some thousands of experiments I feel sure t h a t this will not be of t h e slightest trouble when operations are based upon scientific principles. If we use electrical heating, t h e oxidation is absolutely avoided in t h e furnace shown above in Fig. 8, which was designed a n d built for this very purpose. M y experience leads me t o believe t h a t it can be entirely avoided in fuel-heated furnaces. For example, t h e temperature-concentration diagram in Fig. 13

247

then we could have 2 5 per cent of COZpresent and still have oxidation impossible; t h a t is, we could burn carbon according t o t h e equation, 4c

+

z1/202

= 3CO

+ COZ + 184,830 calories,

without any possibility of oxidizing t h e tube and then the remaining 203,010 calories could be obtained by admitting enough air in t h e preheating zone t o burn t h e 3CO where oxidation would no longer take place because of t h e lower temperature. We might also combine t h e principle of t h e regenerative oven here if we desired. These figures would be modified somewhat by considering the diluting influence of nitrogen, but I have only time t o state general principles without giving details. It is very likely indeed t h a t much more heat may be liberated in the heat zone because of t h e protective influence of a layer of oxide on t h e surface of t h e tube, b u t I always prefer t o calculate on t h e most unfavorable case. Also, we might consider admitting hydrogen with t h e producer gas as by having a steam-blown producer. The hydrogen present would then steadily diffuse through t h e red-hot iron walls as explained above under Fig. I. The outward diffusing hydrogen, aided by t h e outward diffusing carbon a n d carbon monoxide, would meet t h e ingoing layer of oxide a n d would drive it back under t h e right conditions. Also copper tubes would absolutely avoid this oxidation and they would easily stand t h e temperature. I have, in fact, used copper tubes largely a t 950 t o 1000’ in another line of work with t h e greatest success. I could notice no damage t o t h e t u b e either inside or outside. The expense of t h e copper would not be prohibitive and it could be cut down by using it only in t h e actual heat zone, a n d by using i t a s a sheathing for iron tubes; perhaps iron plated with copper would do. I have also tried nickel sheathing with results which would warrant a longer test. There is no doubt t h a t t h e trouble from t h e oxidation of iron tubes has been grossly exaggerated, and, personally, I feel convinced t h a t there is nothing t o it. T H E O R Y O F BATCH F U R N A C E

shows t h a t iron oxide cannot exist in the field above the curve ABC and, hence, i t is impossible even t o oxidize the iron pipe until t h e heating flame contains the corresponding value of C 0 2 . For instance, if t h e iron pipe is heated t o 950’,

The theory of t h e batch furnace is also t h e theory of the process in its simplest form. We merely grind together intimately, iron, coke and soda ash, moisten and briquette t h e material. The dry briquettes are then heated in a vertical pipe, t o say 950’ C., in a current of nitrogen or producer gas until carbon monoxide no longer escapes. This is exceedingly simple and is all t h a t there is t o t h e process. The theory is t h a t the iron, by dissolving t h e carbon, gives a solid solution in which t h e carbon may move freely and i t becomes active. This gives us our “catalytic solution surface.’’ It is hence only necessary t o produce t h e extremely active catalytic solution surface, t o maintain it unimpaired during t h e process, and t o give t h e nitrogen and alkali metal or compound free access t o the surface during t h e run.

248

T H E JOURiVAL O F I N D U S T R I A L A N D E N G I N E E R I N G C H E M I S T R Y E K R I C H I K G P R O D U C E R GAS

You will note from Equation 8

t h a t nitrogen gas is absorbed in t h e process a n d carbon monoxide is formed in large volume. The gas is enriched in a two-fold way a n d I have noted 7 5 per cent of carbon monoxide in t h e exit gas from t h e batch furnaces. Bunsen a n d Playfair absorbed all their nitrogen; hence i t is probable t h a t under t h e right conditions producer gas can be enriched so t h a t i t will be substantially pure carbon monoxide. DESTRUCTIVE E F F E C T OF CARBON MONOXIDE

You will notice t h a t in using producer gas, I always s h u t off t h e gas current before t h e briquettes began t o cool. This is a vital point. For example, briquettes containing 19 per cent of sodium cyanide were heated red hot in a current of producer gas and cooled slowly

Vol. 9 , No. 3

If this be so, then a t joo t o 600' C.,cyanide should be destroyed rapidly (see t h e temperature concentration diagram, Fig. 13) while, at, say, 9jo t o 1000' C., t h e COS has almost disappeared a n d t h e speed of t h e cyanide formation has increased so t h a t no harm is done. This is very important in t h e continuous furnace shown in Fig. 14. THEORY

O F T H E CONTIKCOCS FURNACE

It would take a separate paper t o deal with t h e general features of such a furnace as t h a t shown in Fig. j. The briquettes passing in t h e t o p become preheated, by utilizing t h e waste heat from t h e gases of the reaction zone, and then collect all t h e cyanide fumes which are liable t o distil in clouds from t h e batch furnace. When they get red hot they will, by catalysis of their iron content, produce t h e reaction 2CO

C

+ CO2 + 38,080 calories.

This gives finely divided ash-free carbon which is necessarily i n intimate mixture nyith t h e finely divided iron. I t also gives out heat in t h e briquettes themselves a n d therefore has t h e same exceedingly important effect as internal electric heating. Unfortunately t h e equilibrium relations of curve A B C in Fig. 13 must be applied, but, even if we t a k e the maximum value a t 680' C. this would contribute over zo,ooo calories towards t h e 138,joo calories needed in t h e heat zone for t h e reaction Nz 2Na?C03 4C = 2NaCN 3CO - 138,500 calories.

+

+

+

This is a n exceedingly great advantage. Then in t h e cyanizing zone t h e same reaction takes place as in a batch furnace, b u t t h e intense volatilization from t h e porous briquettes carries t h e alkali upwards t o be recondensed on t h e cooler briquettes above with corresponding heat changes. This distillation gives fine circulation of alkali t o t h e solution surface and is carried back by t h e descending briquette column.

FIG.14

while t h e current continued. When cold, they contained only 3 per cent of cyanide of sodium. This shows t h a t we have almost complete destruction of cyanide when t h e briquettes cool in a producer gas current and t h a t we avoid all trouble in t h e above batch furnace, Fig. 8, b y merely stopping t h e current before t h e charge cools, or, b y dumping t h e hot briquettes so t h a t t h e gas current can not pass through while t h e y are cooling. This destruction may well be due t o carbon dioxide formed by t h e catalysis of t h e iron in causing t h e formation of ashless carbon by t h e equation zC0 C C o n . If this be t h e t r u e explanation, t h e iron acts both as a destructive a n d constructive catalyzer in t h e process.

+

Below t h e cyanizing zone, t h e nitrogen cools t h e briquettes a n d itself becomes preheated so as t o carry t h e waste heat back into t h e heat zone. The plastic briquettes now harden again and are removed by t h e conveyor. When producer gas washed with caustic alkali t o remove C 0 2 is used instead of nitrogen we have in some ways a very different problem. For example, in a run in a 14 ft. 6-in. pipe which should have yielded at least 2 0 per cent of XaCK in t h e briquettes, I got only j per cent of cyanide when feeding a t t h e rate of 2 0 0 0 lbs. per day. When t h e feeding was reduced t o 1000 lbs. per day t h e yield dropped t o I per cent of N a C N in t h e briquettes. I n view of t h e facts already disclosed it is evident t h a t t h e producer gas destroyed "4 of t h e cyanide formed with t h e z,ooo-lb. feed, and when this was made slower t h e producer gas had a longer time t o act a n d 1 9 / 2 ~ of t h e cyanide was destroyed. This looked bad for t h e continuous process with t h e producer gas. The difficulty was overcome, however, b y simply passing t h e producer gas into t h e briquette

Mar.. 1917

T H E J O L - R N A L O F I S D I ’ S T R I A L A N D EA1GI:VEERIA17GC H E M I S T R Y

column a t the base of the heat zone. I n this case the cyanized briquettes as they cool are entirely away from t h e current of producer gas (there must be no leak in the lower fitting of the furnace pipe, otherwise some of t h e producer gas would be forced downvard with the briquettes and thus do harm). This arrangement is shown diagrammatically in Fig. 1 3 and is arranged so t h a t we may use either nitrogen or producer gas and it shows the principle of transferring carbon from one part of a closed system t o another with oxides of carbon so as t o separate from ash; incidentally such a simple modification in the furnace gives 25 per cent S a C N in t h e briquettes. Even poor briquettes will give good results in these continuous furnaces. Some of the advantages can, however, be attained in batch furnaces by running t h e waste gases’and fumes from the top of one into the bottom of t h e next furnace.

249

This reaction takes place TI-ith great ease, hydrogen being given off in large quantity. &\fter agitating for several hours, the hot mass is filtered and the black powder turned back into the process again. The hot filtrate deposits a n exceptionally fine quality of sodium ferrocyanide on cooling, and upon concentrating the filtrate the rest deposits out froin t h e strong solution of caustic soda which is formed. The caustic soda is either turned right back into the process or used otherwise. The method is cscecdingly effective. LIXIVIATIOS FOR CYASIDES

From this behavior of the briquettes with h o t water, we see t h a t it will be necessary t o observe certain precautions in lixiviating t o get sodium cyanide. On t h e other hand, t h e solubility curve in Fig. I j shows t h a t a t temperatures below 3 j O C. the mass will take up mater of crystallization t o form NaCN.2H20 which will “set” the mass just like plaster of Paris.

LIUITIKG FACTS

Suppose we consider, for example, a vertical 6-in. iron pipe, 6 ft. in length charged with briquettes of the usual size. Thus the speed of operation would be limited by the activity of the briquettes, by t h e rate a t which nitrogen could be passed in, and by the r a t e a t which heat could penetrate through the charge t o t h e center of the core. Results already mentioned show t h a t t h e briquettes are active enough t o be cyanized in I or z minutes. The requisite nitrogen could be passed through t h e m in I O minutes. T h e heating could be done in z hrs. without undue excess of temperature outside. These d a t a show why heat conduction is the slow factor and why internal heating b y electricity or by the ashless carbon method are so interesting. a l s o the slow conduction determines t h e furnace design. It means above all else a long heat zone, just as long as possible, so t h a t i t will not be necessary t o force a n excessive amount of heat through a short heat zone with all the ruinous effects t h a t follow. I should use the full length commercial pipes of 2 0 t o z3-ft. length unless t h e plastic briquettes compacted t o o much. I n t h a t case I should t r y slanting the pipes and if this did not work I nrould cut the pipe so as t o bring the preheating zone in a t an angle which should be sure t o reliere t h e pressure. S 0 D I U 11 P E R R 0 C T A SID E

Ha\-ing now s h o n n how t o fix nitrogen in the briquettes, it is in order t o show how compounds can be recovered from them. If v e add hot water t o t h e cyanized briquettes t o make a stiff paste and then steam t h e m in a n agitator you will see a t once t h a t we have a n electrolyte of sodium cyanide and innumerable carbon-iron elements in contact with it. We might hence expect t o get as a result of electrical action, or otherwise, the following reaction: 6NaCN

+ Fe + 2H20 = NaaFe(CN)a

+ 2NaOH + H2.

“1 IO

-20-10

10

0

30 40 5 0 60 70 80 90 100

20

TEMPERATURE IN OC Fro. 15

To avoid both dilemmas we should lixiviate a t a ternperature slightly above 3 j o C., the transition point of the cyanide. This should be done quickly because the ferrocyanide reaction is going on slowly even a t this temperature. The solution can be concentrated i i t vacuo if desired. The cyanide is pure enough for many purposes. Or, i t can be distilled as mentioned above. I have sometimes added lime t o it t o causticize any soda ash present as this would prevent the hydrolytic dissociation t o form hydrocyanic acid, SaCN

+ H 2 0I_ S a O H + H C S ,

for vhich manufacturers frequently add caustic soda intentionally. Also we may add calcium chloride t o remove the carbonate and replace i t by sodium chloride thus : Na2C03

+ CaCh

=

C a C 0 3 $- 2NaC1.

On then evaporating, we have the mixture of sodium

= 50

T H E J O U R N A L OF I N D U S T R I A L A N D ENGINEERING C H E M I S T R Y

chloride and sodium cyanide which is so much sought in the jewelry trade. AMMONIA A N D SODIUM FORMATE

If we boil sodium cyanide solution t o get ammonia a n d sodium formate, NaCN

+ 2H20

= HCOzNa

+ NH3,

we can only get t o the boiling point of t h e solution. t h e action is slow and hydrocyanic acid is likely t o form according t o t h e already mentioned hydrolytic dissociation.

If, however, we add caustic soda (or prepare i t with lime in t h e lixiviated solution), we get these effects: t h e caustic soda prevents hydrolytic dissociation t o form hydrocyanic acid, i t raises t h e boiling-point of t h e solution and it salts out t h e sodium formate from t h e solution by making i t many times less soluble t h a n in pure water. T h e operation is carried out with a return condenser a n d the ammonia after passing through a caustic soda dryer is chemically pure a n d anhydrous and can be compressed a t once into cylinders, if desired. The sodium formate can be turned back into t h e cyanizing process or it can be converted into sodium oxalate a t once by merely heating in vacuo; from these we can easily obtain formic and oxalic acids. U R E A , AMMONIA A N D NITROGEN

If we can arrange a flanged cast-iron pot, 3 in. in diameter b y 1 2 in. deep, with a cover, inlet a n d

Vol. 9 , No. 3

cyanide into urea. The cyanide is kept just above its melting point and a current of air is passed through t h e molten cyanide. It burns t o sodium cyanate according t o the approximate equation,

+ Air

NaCN

= NaCNO

+ zNZ.

The nitrogen thus formed is turned into t h e cyanizing furnaces and this gives us 4 atoms of nitrogen for each 2 atoms fixed. The process hence takes its own nitrogen from t h e air t o such a n extent t h a t theoretically one-half would have t o be thrown away as a waste product in spite of t h e process being one for t h e fixation of nitrogen. One-half of the cyanate is now heated with water which converts it into ammonia long before t h e boiling point is reached according t o t h e equation, NaCNO

+ zHzO = N a H C 0 3 + NHs.

The sodium bicarbonate is thrown back into t h e process and t h e ammonia saved or else converted into urea as follows. The other half of the NaCNO is added t o water and t h e ammonia then passed in. Then COz from t h e cyanizing furnaces is also passed into t h e liquid. The change is represented by, NaCNO

+ NH3 + H20 + COZ + NH4CNO and NHdCNO

= NaHC03

= CO(NHz)S (Urea).

You will notice t h a t this is exactly similar t o the ammonia-soda process, NaCl

+ NH3 + HzO + COz

NaHC03

+ NH4Cl,-

t h e radical -CNO taking t h e place of t h e chlorine atom of sodium chloride. The urea reaction takes place much better t h a n t h e ammonia-soda one because t h e "&NO is passed into urea and hence practically does away with reversibility while t h e NHdCl in t h e ammonia-soda process causes a loss of 30 or 4 0 per cent because of this factor. The bicarbonate is returned t o t h e process. We thus have a very inexpensive way of getting urea which has over 46 per cent of nitrogen and which is about three times as rich in nitrogen as sodium nitrate. It is about twice as rich in nitrogen as ammonium sulfate and would not introduce sulfuric acid into our soils, many of which are already too acid. The urea was tested in water culture in t h e botanical laboratory of this university a n d found t o give results equal t o t h a t of t h e nitrogen from potassium nitrate. The urea may also be combined with nitric acid produced from ammonia so as t o form urea nitrate ( c o ( N H ~ ) ~ . H N o ~ ,which is still very rich in nitrogen b u t less soluble t h a n urea. Fro. 16

exit pipes, and a 2-in. reducing coupling as shown in Fig. 16, we have a serviceable laboratory unit t o convert

METALLIC SODIUM

Sodium cyanide melts a t a much lower temperature t h a n salt and requires much less voltage for electrol-

Mar., 1917

T H E J O U R N A L OF I N D U S T R I A L A N D ENGIiVEERING C H E M I S T R Y

ysis. My experiments indicate t h a t it can be easily electrolyzed [2NaCN = 2Na (CN)z] t o sodium and cyanogen. This should give us a very cheap way of getting sodium from sodium carbonate with t h e incidental fixation of nitrogen.

+

OXAMID, OXALIC A N D FORMIC ACIDS

The cyanogen obtained from above was not supposed t o be readily convertible into oxamid b u t I find it converts with t h e utmost ease under t h e right conditions. The gas is rapidly absorbed in hydrochloric acid (44 per cent) and then changes t o oxamid thus: (CN)z zH20 = CO(NHz)*, which being practically insoluble separates out as a pure white powder.

+

Here we’have t h e remarkable case of t h e hydrochloric acid acting as a catalyzer t o concentrate itself even t o t h e point of becoming gaseous, while a t the same time precipitating out t h e resultant compound, Curiously enough, t h e reaction practically stops when the hydrochloric acid is slightly dilute (14 per cent). The oxamid has nearly 3 2 per cent of nitrogen and is nearly insoluble in water and i t should be especially of value as a fertilizer and should partially approximate the effect of t h e nitrogen in dried blood rather t h a n t h a t from t h e more soluble sodium nitrate. This has apparently already been verified by Dr. Hartwell of t h e Rhode Island Agricultural Experiment Station. We get chemically pure oxalic acid from t h e oxamid by merely heating i t for a minute or so with t h e concentrated hydrochloric acid t o add more water, thus: (CONH2),

+ 4 H z 0 + 2HCl = (COOH)z.zHzO

The acid crystallizes salts.

out

free

from

+ 2KH4C1. all

mineral

The acid upon heating with glycerin transforms a t once into formic acid according t o t h e equation, (COOH), = HCOzH COz.

+

CAUSES OF F A I L U R E

A few of t h e large number of reasons for t h e absolute failure t o achieve commercial results worthy of consideration by t h e methods described in this paper may be given: ( I ) The omission of iron (which, by t h e way, is probably t h e only technically suitable catalyzer where alkali carbonates are used), absolutely ruins t h e process. ( 2 ) Failure t o mix the constituent properly is nearly equally ruinous and has no doubt contributed largely t o failure.

( 3 ) Heating a briquette a little too high will ‘melt the iron a t t h e eutectic point into globules, thus destroying t h e extended catalytic solution surface and thereby completely ruining t h e process.

I t is also above shown t h a t simply putting a large vertical retort in the furnace horizontally will (4)

2

jI

completely ruin the process whenever t h e charge is plastic. Alder preferred t o mix his powdered materials with water and then with pieces of charcoal from “size of pea t o size of fist.’’ You can see a t once t h a t here t h e charcoal would absorb t h e molten cyanide just as a sponge absorbs water, thereby preventing t h e alkali from contacting freely with the catalytic solution surface even when he had one present by using iron. T o show how serious this is, I prepared an active ironcoke-soda ash mixture according t o my specifications which yielded cyanide of 98 per cent purity. A portion was then mixed with water and charcoal and i t now yielded cyanide of only 14 per cent purity. This shows conclusively t h a t Alder’s preferred directions will ruin even a n initially active mixture. You can now see why I was c’areful t o have the screens spaced in Fig. 1 1 t o keep t h e coke and briquettes apart. ( 5 ) The above work shows t h a t cooling a cyanized charge containing iron in producer gas ruins the process. (6) Evidently oxygen would be worse and you can form your own inference about cooling t h e descending charge in nitrogen containing 2 . 5 per cent of oxygen. ( 7 ) The quantitative measurement of heat penetration also shows t h a t t h e heat zone should be long so as t o allow time for t h e Beat t o penetrate t o t h e center of t h e charge without having t o force t h e process and thereby causing no end of needless trouble. EMERGENCY PROCESS

I have carefully distinguished between what I have actually demonstrated with units which were of commercial size and those other things which have not been worked out as far. I n view of t h e conditions now existing, I wish t o call attention t o this topic so definitely t h a t t h e issue will be so sharply defined t h a t anyone interested can test the matter for himself by direct experiment within a few hours. This is a perfectly fair way of finding t h e t r u t h or falsity of t h e statements and criticisms I have felt compelled t o make.

If one wishes a n operative process a t once, it is only necessary t o construct t h e furnace shown in Fig. 8. Construct a platform about 6 feet by 8 feet with 6 feet of free space underneath out of 4 X 6 timbers and 2-in. planks, cutting a hole through which t h e iron pipe may pass. P u t on four small piles of brick t o support t h e metal drum which should be, say, 30 in. in diameter by 6 feet in height and fill i t with magnesia asbestos after inserting t h e drilled pipe and fastening with locknuts. P u t on t h e reducing coupling, stuffing box and iron support as shown in Fig. 8 and drop in t h e brick support 8a and turn on the 8-in. receiving pipe. This completes the furnace and with t h e help of several workmen, I would find no trouble in doing t h e whole thing in 6 hours. Briquettes are now added and

T H E J O U R N A L O F I N D U S T R I A L A N D ENGINEERIAVG C H E M I S T R Y

2j 2

t h e copper leads attached. With sufficient current t h e first run will be finished a n d drawn within 2 hours and t h e n runs can be made every hour; this can be kept u p indefinitely if t h e drum be air-tight. This furnace can be in operation in less t h a n 2 4 hours, giving its output steadily and will easily yield Ijo lbs. of sodium cyanide per 2 4 hours or over 2 0 0 lbs. of sodium ferrocyanide in t h e same time. An ordinary meat chopper will easily supply briquettes enough t o run 8 such furnaces. With over $2.00 per lb. now asked for sodium cyanide, a n d with sodium ferrocyanide a t over $1.25 per lb. a n d potassium ferrocyanide at over $7.00 per lb. within t h e past I O months a n d great anxiety i n both t h e mining a n d jewelry industries i t seems t h a t t h e above furnace which costs less t h a n $100.00 and can be p u t u p and operating in less t h a n a d a y might have been tried.

It would, at a stroke, either have shown how t o relieve t h e situation immediately or else have shown t h a t what I have been telling you to-night is worthless.

If no high amperage is available, t h e ordinary lighting current of 1 2 0 or 2 2 0 volts with iron rods analogous t o t h e construction above indicated in Fig. g would work just as well. If electricity were not available t h e n a n oil burner would answer just as well as a n y burningout of pipes would not need t o be considered a t such prices. I a m not here advocating a n y particular furnace, hence there is no cause for discussion about its merit or lack of merit. It simply shows a means of getting away from endless discussion b y doing something t o get chemicals which are sorely needed. These units could be enlarged a n d p u t in series t o take practically a n y current and I have d a t a for t h e design of a plant capable of producing t h e ammonia for 180,000 tons of nitric acid per year along these lines and it would be something which could be done quickly in case of need. COST O F PROCESS

I do not have time t o go into t h e question of cost of production this evening b u t I have estimated i t very carefully from what I have actually tested with commercial units as vel1 as from theory. I will hence say t h a t t h e equation, Ka2COa

+ 4C + S2 =

I

2SaCX

+ gCO

-

1 3 S , ~ o ocalories,

x o u l d theoretically require about 3 j , o o o H. P. t o produce 180.000 tons of nitric acid, allowing 8; per cent efficiency in oxidizing t h e ammonia. But t h e three molecules of CO in t h e above equations would give on burning, gCO

+ 1 ~ / * 0= ?3C02+

200,000

calories,

equivalent t o about jo,ooo H. P. t o help make up t h e unavoidable heat losses. Combining these t w o equations we have

Tol. g 3 S o . 3

+ 4 C + 1 ~ /+~N20 ~

Na2C03

=

zNaCN

+ 3 C 0 2 + 61,500 calories.

The total process is hence really a n exothermic one and, in any case, i t does not require electric power. Owing t o its great simplicity and t o t h e inexpensive materials, t h e cost is exceptionally favorable for t h e fixation of nitrogen and t h e production of cyanides, ammonia, urea a n d t h e other substances above mentioned. C 0 M P A R I S 0S S

We are now in a position t o compare t h e above process with three of t h e commercial methods now in successful operation. The arc process which has been developed on such a large scale in Norway yields nitric acid and nitrates but i t requires a very high temperature a n d is hence absolutely dependent not only on electric power, but, I think all agree t h a t i t can succeed only where t h e power is very cheap because t h e requirement of current is very large. It is further handicapped by yielding very dilute products which must be worked over. The Cyanamid process requires calcium carbide which can be made practically only at t h e high temperatures of t h e electric arc. It is therefore also absolutely dependent upon electric power b u t i t can make use of more expensive power t h a n t h e arc process.

It has hence a correspondingly wider choice of locations. I t is also handicapped by using nitrogen itself rather t h a n producer gas. M y process differs from the above processes by operating a t temperatures which may be below 950' C. and it is therefore independent of electric power. This means t h a t it can be operated at a n y locality suitable for manufacturing purposes. I can use producer gas just as well as nitrogen and t h e process is therefore independent of pure atmospheric nitrogen. I t is so exceedingly simple t h a t i t can be installed immediately with apparatus obtainable in t h e open market and because of t h e remarkable transformations of t h e cyanogen it opens t h e way for t h e cheap production of organic substances such as urea, oxamid, etc., as well as new commercial methods for oxalic and formic acids and many other substances I have in mind. The Haber process fixes nitrogen by combining it with hydrogen and requires these substances in the pure state. I t requires costly machinery, and its development required t h e highest t y p e of engineering skill. These highly specialized things, horn-ever, mean t h a t i t will also require especially competent people t o run it. On the other hand, t h e process I have described t o you this evening represents t h e other extreme. I t makes use of the crudest things such as coke, producer gas a n d water. These are handled by t h e simplest possible mechanical operations and t h e catalyzer, iron, is not bothered even by t h e gross assortment of elements in t h e ash of t h e coke, and any sort of labor, when properly directed, can handle it.

? \ I a r , 191;

D E N G I iY E E R I S G C H E M I S T R Y T H E JO17R,VAL O F I S D I ’ S T R I A L A *IT REMARKS

I t is not worth while t o make a summary of the work I have described, because the whole presentation has been condensed so as practically t o be a summary. Almost everything mentioned can be done in a variety of ways which I did not have time t o mention and even the most important things could not be much more t h a n mentioned. I hope t h a t t h e evidence presented to-night has cleared away some of the erroneous statements which have hindered progress in this field which has been handled in a n esceptionally unfortunate way. I also hope t h a t the statements under “Emergency Process” will, if sudden need arise, lead t o instant action instead of interminable discussion and counter-propositions. The way t o test t h e matter is t o set up t h e apparatus as is sharply set forth. A roof can be put over i t after i t is in operation and blueprints, plans and improvements can await their turn. S o n e of the things mentioned in the above work were discovered by chance. All were predicted b y a carefully considered application of t h e very simple but fundamental principles of physics, chemistry and mathematics, which every undergraduate in chemistry is taught. They Tvere then verified b y experiments which were chosen t o give decisive answers t o questions and t o establish quantitative data. Only a comparatively short time could be given t o the work because of the heavy pressure of other duties, and it would certainly have been impossible t o carry i t out if it Tvere not true t h a t pure science is the only foundation for industrial work. I have been connected with universities and technical schools for 28 years as student and teacher. I haT-e alnxys used every effort t o further the view t h a t all industries must be based directly on work in pure science and t h a t the only way t o succeed is t o become thoroughly master of these fundamental principles and then look a t things from eyery point of view in order t o solve whatever problems may be encountered. There seems t o be a constantly increasing pressure both from within as well as from without the university. which is inimical t o this ideal and many cures, which are at varience l\rith i t , are advocated, in the rush t o get results quickly by a short cut t o knowledge. If i t had been more generally realized t h a t hard work and careful, self-sacrificing preparation are very likely ultimately t o lead t o quick, certain, solution of difficulties rather than perhaps t o years of misguided effort and enormous expenditures of money, with, like as not, ultimate failure, things would not have drifted t o their present condition. If these illustrations have some influence in this direction I believe t h a t i t will be more important t o the country t h a n even a successful nitrogen fixation process. BRONVN UNIVERSITY

R. I. PROVIDENCE,

2

53

THE VOLATILIZATION OF POTASH FROM CEMENT MATERIALS‘ BY E. -4NDERSON

AND R. J. NESTELL Received January 25, 1917

I n the course of a n extended series of investigations upon t h e subject of potash volatilization from silicate mixtures, the following report was prepared, covering t h a t portion of the laboratory work which was incident t o the study of volatilization of potash from cement making materials. T h e work reviewed in this report forms one link in a chain of investigations covering various possibilities of volatilizing potash from silicates during t h e process of manufacturing Portland cement, with subsequent collection of the volatilized potash from t h e furnace gases by means of electrical precipitation. The process of electrical precipitation can be applied in two ways: should the total dust and fume issuing from t h e furnace contain a large percentage of potash. t h e total suspended material can be collected in n single electrical precipitator; should, on the other hand, the potash content be low, a fractionation of the suspended material may be resorted t o , as for e r ample b y utilizing a tvo-stage fractional precipitation apparatus. OBJECT-It has long been known t h a t the raw material used in cement making contains potassium salts which are partly volatilized in the burning, and from experiments in field and laboratory, some d a t a bearing on this volatilization have been obtained and published. Very little exact information, and particularly such directly applicable t o the potash volatilization from actual cement materials has. however, been available, and t h e following series of experiments were undcrtaken for the purpose of obtaining such information regarding the factors which influence the liberation of potash in actual cement burning. S C O P E O F T H E ISvESTIGATIOS-The rate O f volatilization of a potash salt from cement mix in the highly heated zone of a cement kiln, is dependent upon a number of factors. -4s predominating factors affecting the possible recovery in the furnace gases beyond the furnace, there may be mentioned: ( I ) the temperature prevailing in the kiln; ( 2 ) volume of gas passing; (3) the intimacy of contact between the furnace gases and t h e cement mix; (4) the vapor pressure of t h e potash salt or salts formed; (j) the possibility of dissociation under certain furnace conditions (oxidizing, neutral or reducing atmosphere or changing temperature) t o components of greater or less volatility t h a n the original salt; (6) the degree of saturation of the gas in contact with cement material; ( 7 ) the rate of diffusion both of the salt vaporizing in the interstices of t h e cement mix t o the surface of contact with the gas stream, and of the saturated gas a t the surface t o the leaner gas areas beyond. Furthermore, since in cement burning, cement material changes from a mixture of lime and clay t o a clinker consisting largely of silicates and aluminates of calcium, the release or formation of t h e potash salt 1 Report on Laboratory Investigation by the Western Precipitation Company.