December, 1946
INDUSTRIAL AND ENGINEERING CHEMISTRY
which is comparable with the highest value, 1.60, observed in these experiments. Similar calculations for wool, for a normal specific weight of 1.30, indicate 1.38 a t 20,000 pounds compared with an observed maximum of 1.35, This hypothesis was not considered further in developing the relations given in this paper, because there are too many assumed values involved and because the equation which has been developed is sufficiently accurate without this further refinement.
1315
cylinder). At lower pressures small samples are denser than large samples. The effect of liquid on the ressure-volume relation depends on whether the fiber is affectei chemically or physically by the liauid. If there is no reaction the P - ' I relation is verv close to t h i t of dry fiber. If the fiber is hydrophilic the presence of water causes higher density a t low pressures, compared to dry samples. The effects of other variables were considered in .this report, but they are of lesser importance than those mentioned or else the data are not sufficiently detailed t o warrant further conclusions.
CONCLUSIONS
A logical first step in the development of expression theory is the formulation of a relation between the volume of a n expressible material and the pressure upon it. For the series of fibrous materials studied in this paper and in previous work by the present writers and others, the following relation is found valid for most cases : P = ab('/v) log P = a' b'lV
+
This equation gives a close fit to observed data for experiments on cotton, wool, paper pulp, felt, sawdust, asbestos, sugar cane, and bagasse. It is not valid a t low pressures (below about 50 pounds.per square inch) or in cylinders of small diameter (such a? Deerr's '/r-inch Cylinder) but seems generally applicable otherwise. * The effect of sample size is negligible a t pressures greater than 500 t o 1000 pounds per square inch (in the ll/s-inch diameter
LITERATURE CITED
(1) Bridgman, P. W.,personal communication, 1941. (2) Deerr, N.,"Cane Sugar", London, Winthrop Rodgers, Ltd., 1921. (3) Deerr, N.,Hawaiian Sugar Planters' Assoc., Bull. 22 (1908). (4) Ibid., 30 (1910); Arch. Suikerind., 19,21-60 (1911). (5) Deerr, N., Hawaiian Sugar Planters' Assoc., Bull. 38 (1912); Intern. Sugar J., 14,lS(1912-13). (6) Koo, E.C., IND.ENQ.CHEM.,34,342-5(1942). (7) Koo, E. C., J . Chem. Eng. China,4,15-20,207-11 (1937);5,4752,69-73(1938);7,1-4,23-5(1940); 8,l-10(1941). (8) Koo, E. C., and Chen, S. M., Ind. Research (China), 6, 9-14 (1937). (9) Wright, J. W., and Bennett, C. A., U. S. Bur. Agr. Chem. & Eng., Mimeo. Pub. ACE-67 (1940). SUBMITTPD in partial fulfillment of the requirements of the degree of doctor of engineering science, College of Engineering, New York University.
Improved Preparation of Guanidine Nitrate K. G. HERRING, L. E. TOOMBS, R. S. STUART, AND GEORGE F WRIGHT University of Toronto, Toronto, Ontario
A new process is describedby which guanidine nitrate can be obtained in 92449% yield from lime nitrogen, ammonium nitrate, and urea. The latter compound is carried repeatedly through the process, together with excess ammonium nitrate and some dissolved guanidine nitrate, and serves to keep the reaction mixture fluid at the temperature (120" C.) at which the conversion occurs. This anhydrous reaction medium reduces by-product formation, reactor corrosion, and difficulty with wet ammonia. The comparatively low reaction temperature obviates the necessity for pressure equipment and reduces hazard in manipulation of ammonium nitrate containing organic material.
T
H E chemical literature records three main sources*ofguanidine salts-namely, from dicyandiamide, from cyanamide solutions, and from cyanamide salts. The first of these, dicyandiamide, does not seem to have been used as a principal source of guanidine nitrate before 1908, when a German patent (22) of that date described its fusion with ammonium nitrate to produce the guanidine salt. This was generalized in 1912 (1) and then extended by Werner and Bell in 1915 (27) and later in 1920 (28). These workers, using ammonium thiocyanate, considered that, since dicyandiamide is converted a t 205" C. to cyanamide, this depolymerization constitutes the first step of the reaction in which cyanamide is subsequently ammoniated to guanidine. Davis (6-9) described and patented a similar anhydrous fusion process, which was also reported by Ewan and Young (II), Blair and Braham (S),and Kat6 et al. (16). The process works poorly below 160" C.; above 170" it proceeds exothermically but not explosively. Infusible by-products were always reported, to-
gether with guanidine nitrate. This work, especially that of Davis, was reviewed in 1931 ($0). The Davis mechanism involves initial reaction t o form biguanide (the main product below 160' C.), which then would ammonolyze t o form guanidine nitrate.
/"
N
H
/H
N
H
NHlNOI
N
$
/" 160" C.
One may suspect, after carrying out this fusion in the laboratory, that it would be somewhat hazardous on a large scale. FurthermoPe, Davis (6)implies that the 80% yield obtained under anhydrous conditions can be maintained if water is added, but Blair and Braham (8)report a more modest 54% yield of guanidine nitrate together with large amounts of biguanide nitrate and unchanged dicyandiamide, when the latter substance is treated with aqueous ammonium nitrate at 170-180" C. under 50pound pressure for 3 hours. Our results are in exact agreement with those of Blair and Braham. Thus, while water or ethanol always seem to lead to incomplete conversion and to by-product formation, the use of a n anhydrous diluent-liquid ammonia-has been used successfully (14) by the American Cyanamid Company to give a clean product in good yield by a method relatively free from the danger inherent in the
1316
INDUSTRIAL AND ENGINEERING CHEMISTRY
exothermic dry fusion process. By contrast, however, corrosion may be a factor, and pressure equipment is required. All of the processes described employ dicyandiamide. Since this compound is usually prepared by alkaline dimerization of cyanamide solutions, the simplification has been proposed by some workers that cyanamide solution be used directly as a guanidine Hource. Thus Erlenmeyer (IO) and Volhard ($4) obtained guanidine salts by heating water or ethanol solutions of cyanamide with ammonium chloride or thiocyanate a t 100" C. for 10 hours. Schmidt (19) carried out the same reaction with ammonium hydroxide and obtained much dicyandiamide with his guanidine, whereas Ewan and Young ( 1 1 ) reported, in 1921, a 4370 yield of guanidine nitrate by neutralizing a concentrated aqueous solution of sodium cyanamide with nitric acid and then heating with ammonium nitrate at 150" C. The latter process w s improved by Blair and Braham ( 2 , 4) by heating 180 grams of cyanamide per liter of solution with a 27% excess of ammonium nitrate in an autoclave at 155" C. for 2 hours. A discussion of the several variables as they affcrt the optimum S O Y , yield of guanidine nitrate is given in detail. In particular these authors postulate three competing reactions -ammoniation (reaction l), hydration (reaction 2), and poly,mcrization (reaction 3):
H
/
N
+
m n x----f NH2-k-kH2.HX H&CN + H2O +NHsCOSHz
H2Ecx
;
(1) (2)
/H
OH-
2H2SCN -+H2S-
-KHCN
(3)
Reaction 3 could be minimized at temperatures above 160" C. Reaction 2 consistently accounted for about 10% of the cyanamide, whereas 8% of the Cyanamide appeared as infusible products. A similar process by Jones and Aldred (15) yielded 70% of guanidine phosphate when one part of cyanamide, 2 parts of ammonia, and 1 part of phosphoric acid were autoclaved in aqueous solution at 140" c. This process involves the use of free cyanamide, but others reported a desirable modification by use of the sodium or calcium salt, which eliminates the separate conversion to cyanamide. Thus Imperial Chemical Industries (G) claim a 90% yicld of guanidine nitrabe from aqueous ammonium nitrate and the salts of cyanamide at 100-125" C. When the present authors repeated these procedures using lime nitrogen as the cyanamide source, the yields did not exceed 50% of theoretical. Such low yields might have been predicted for reaction in aqueous solution of crude cyanamide containing enough calcium oxide to render the solution thoroughly alkaline. The dicyandiamide produced by this alkalinity could not be converted to guanidine below 160" C. On the other hand, convcrsion to dicyandiamide seems not to be so easy in anhydrous media. A German patent of 1925 (IS) refers to fusion of calcium cyanamide with ammonium nitrate, whereas Gockel ( I $ ) , having first ascertained that ammonia facilitates the reaction, proposed in a patent with Traube (23) that the reaction mixture calcium cyanamide-ammonium nitrate will be liquefied at 90-100" C. by the ammonia produced, first from ammonium nitrate with the lime in crude calcium cyanamide, and later by ammonia freed during the process according to Equation 4:
N-H
CaCNl
/I + ~NH~NOI-NH~-C-NH~.HNO*
+ 2NH1 +
C a W O h (4)
Vol. 38, No. 12
This rcport of a process which is attractive in its simplicity does not stress yields or describe a procedure which would be workable from the cconomic standpoint. A more comprehensive report is presented by Spurlin ($1) who heated calcium cyanamide and ammonium nitrate from 110" to 170" C. The reaction representative of Gockel's process is desirable €ram several aspects. It obviates clumsy extraction of cyanamide and conversion of the latter to dicyandiamide. The low temperature at which it may be carried out reduces corrosion. Certain difficulties arc apparent, however, when the reaction is considered in a practical sense. The fluidity of the reaction mixture which is obtained in Gockel's process is a result of reaction rather than a contribution to it. KO eutectic between ammonium nitrate and calcium cyanamide is involved, but, instead, a solubilization of the latter salt occurs in the Diver's liquid (ammonium nitrate plus ammonia) Yhich is produced according t o reaction 5 :
But Diver's liquid tends to lose ammonia a t temperaturL3 of 90120 C. This is evident if one heats the intimate mixture of lime nitrogen and ammonium nitrate to 120" C. Violent frothing occurs as a result of ammonia evolution, and this becomes less controllable as the depth of initial mixture is increased. When the ammonia evolution has subsided, the fluidity of the mixture varies according to the amount of guanidine nitrate and water formed from the calcium cyanamide and calcium oxide, respectively. Stirring, necess:iry to homogenize this mixture completely, usually is effective only if the temperature is raised from 120" to 160" C. By contrast, Gockel's reaction mixture is finaIIy completely fluid if the mixture is heated to 120" C. in a closed system which retains the ammonia. The pressure involved in this instance does, however, prescnt certain difficulties if the nitrolimc is to be added gradually to avoid the progressively strenuous reaction. Under any circumstances the reaction mixture is initially solid. An alternate process was therefore sought which did not depend on solubilization by ammonia to keep the reaction mixture fiuid. Such a process would ensure easy stirring from the outset and, further, would ensure that fluidity minimize the frothing due to ammonia evolution. Since the ammonia would riot be needed as a solubilizing agent, such a reaction could be carried out at atmospheric pressure. It was found that this could be accomplished by use of the urea-ammonium nitrate magma, which has a eutectic temperature of 49" C. The introduction of urea was also advantageous in reducing the hazard of heating ammonium nitrate with carbonaceous materials. The reaction thus had all the advantages with respect to fluidity and safety which characterized the aqueous procesi of Burns and Gay ( 5 ) and, in our experience, gave a much higher yield. Although mild or cold-rolled steel was corroded by our reaction mixture at a prohibitive rat,e (90-120 mm. yearly bite a t 120"C.), corrosion'was not measurable on 18-8stainless steel at a reaction temperature up to 130" C. At 160" C. the rate of bite was 4 mm. per year. .These corrosion rates were measured on samples including electric welds which had not been heat-treated. A sample which did not include a weld was corroded at a yearly bite of 1 mm. at 160" C. EXPERIMENTAL VARIABLES
The optimum reaction conditions were worked out in stirred, mercury-sealed glass vessels containing 40.0 grams of lime nitrogen ( 6 2 4 % calcium cyanamide, 13-14% calcium oxide). These vessels were vented to ammonia collecting systems. I n order that the several variations listed below would not depend on critical temperature and time ranges, the vessels were first held a t 160-1?0" C. and 2 hours, respectively, and then varied downward with respect to the optimum conditions obtained under these temperature-time maxima.
December, 1946
INDUSTRIAL AND ENGINEERING CHEMISTRY
EFFECTOF EXCESS AMMONIUMNITRATE. I n this series of experiments 40.0 grams of lime nitrogen (62% calcium cyanamide, 13% calcium oxide) were mixed intimately in a mortar with various amounts of ammonium nitrate containing 25% of its weight of urea. The mixture fused a t 85 O C. ; this indicated that the presence of lime nitrogen lowers the melting point of ammonium nitrate-urea below the 100' C. fusion point expected fo: this composition. The temperature was raised quickly t o 160 after completion of the initial reaction, which required about 15 minutes. After 2 hours at 160-170" C. the mixture was cooled to 120" C. and diluted with enough water to dissolve the nitrate and urea. After filtration to remove carbonaceous matter, the guanidine nitrate was precipitated by chilling. An excess of 33% over the 89 grams (3 equivalents to react with the calcium cyanamide and 2 equivalents t o react with calcium oxide) of ammonium nitrate required for the reaction gave an 85% yield of guanidine nitrate; a 50% excess gave a 90% yield, and a 100% excess gave a 95% yield. EFFECTOF UREACONTENT.This variable was evaluated under the same reaction conditions, with 40 grams of lime nitrogen, a 33% excess of a m m o j u m nitrate, and different amounts of urea. When the urea was 15% of the weight of ammonium nitrate, the guanidine nitrate yield was 93% of theoretical. A urea content of 25% lowered this yield t o 91%, and a content of 45% (melting point, 50" C.) decreased the guanidine nitrate yield to 67%. It is evident that the percentage of urea should be held as low as possible, and 15% urea content (meltin point, 120" C.) was maintained in most of the work. The sligtt diminution in yield when 25% urea was used might, however, be compensated by ease of handling of the 100" C. melt. A 33% excess of ammonium niEFFECTOF TEMPERATURE. trate (1 18 grams), containing 25% of its weight of urea, gave 91 % yields of guanidine nitrate when heated with 40 grams of lime nitrogen for 2 hours at 160-170" C. Under otherwise comparable conditions a reaction temperature of 120-130' C. gave a yield of only 63%. This low yield might have been expected in view of the large amount of dicyandiamide which was also present: the latter compound, in the same ammonium nitrate-urea mixture, is converted to guanidine nitrate t o the extent of only 38% in 2 hours. At 160-170" C., on the other hand, this conversion is 87% complete in the same length of time. EFFECTOF ADDITIONRATE. It was evident that a low temperature reaction could not be attained under the reaction conditions imposed in the foregoing experiments. These conditions were, however, objectionable from the standpoint of the vigor of the initial reaction, which foamed badly because of rapid ammonia evolution. The reaction conditions were altered on this account by addition of 40 grams of lime nitrogen to the mixture composed of 118grams of ammonium nitrate and 17grams of urea. The addition was made over 30 minutes with the mixture a t 120" C. After 2 hours more at this temperature the mixture was processed as before to yield 80% as guanidine nitrate. The remainder consisted of a small amount of dicyandiamide and considerable cyanamide. When the aqueous liquor was evaporated t o complete dryness and ammonium nitrate added t o compensate for that converted to guanidine nitrate, the re-use of this melt at 120°C. with 40 grams of fresh lime nitrogen gave a 95% yield, while subsequent re-use of the reaction mixture gave a constant 97% yield of good guanidine nitrate. This yield increase t o a constant value indicated that the cyanamide and dicyandiamide had reached limiting concentrations at which they were consumed during the reaction period. These results might have been anticipated from another series of experiments wherein 40 grams of lime nitrogen were premixed with varying amounts of ammonium nitrate containing 15% of their weights of urea; the mix was then heated t o 120" for 2 hours. This series showed t h a t a 33% excess of ammonium nitrate over the 89 grams required by theory gave a guanidine nitrate yield of 68%, a 100% excess gave a yield of 7574, a 200% excess yielded 90%, and a 300% excess yielded 95%. Slow addition of lime nitrogen would assure'a large proportionate excess of ammonium nitrate at any time and hence would result in high yields when a moderate excess was employed. The action of the excess ammonium nitrate is thus probably that of an extractant which frees the cyanamide and consumes the calcium oxide present in the lime nitrogen, with ammonia freed in the process. Since the free cyanamide is in homogeneous solution with a large excess of ammonium nitrate, the mass effect of the excess component is t o favor guanidine nitrate formation and also minimize dicyandiamide synthesis. EFFECTOF MOISTURE. The amount of moisture introduced by fresh lime nitrogen evidently is negligible, because the condensed ammonia from a typical preparation contained only 0.02% moisture, as was demonstrated by the gas evolved on addition of sodium. It is important that the reaction mixture be essentially
1317
anhydrous. When water t o the exteni of 5% of the weight of 178 grams of ammonium nitrate and 27 grams of urea were added toether with 40 grams of lime nitrogen, the reaction mixture, feated to 160-170' C. for 2 hours, yielded only 87% as guanidine nitrate instead of t h e expected 96%. This demonstrates the importance of distilling the water completely from any ammonium nitrate excesses which are re-used in subsequent reactions.
I
MAKE UP URFA
STORAGE
DRY
NU3
I HOT FILTRATIOU
G
j.
I ' SLUOGC
i
1
ATION
25%.
I
1-4
EVAPORATION
14O'C. AT IOM
t
WATER 2.5 LE.
I
RETURNED
TO REACTION
Figure 1. Flow Sheet of Ammonium Nitrateurea Fusion Method for Production of Guanidine Nitrate
It is possible that small amounts of water react by hydration of cyanamide to urea. If hydration did occur, it must have produced less urea than was lost mechanically or chemically, because urea did not accumulate in the re-used reaction mixture. Urea was not consumed in the principal reaction, although it may have been hydrolyzed t o ammonium carbonate. A blank run containing urea, ammonium nitrate, and calcium oxide yielded no guanidine nitrate. This result might have been expected, because the reported formation of guanidine in this manner (18) does not occur below 250' C. PURITY OF REAGENTS. Two of the reagents, ammonium nitrate and urea, are among the purest of all industrial chemicals and caused no difficulty throughout the study. Only one uniform lot of lime nitrogen was used as supplied by Welland Chemicals Limited. The analysis of this lot was reported by them as follows: Calcium cyanamide Total nitrogen Water insoluble nitrogen Silica and insolublea Ferrio oxide and aluminum oxide Carbon
22.53% 0.9 2.19 1.51 12.12 0.43-0.90 1.76-2.74 1.01 1.19-1 .65 1.06 13.88
- Only the cyanamide content was checked, by the method of Pinck ( I 7 ) , before each experiment. EFFECT OF CATALYSTS. No catalyst was found that accelerated the reaction within the experimental error. Nickel, which forma
1318
INDUSTRIAL AND ENGINEERING CHEMISTRY
an insoluble biguanide salt, did not depress the yield; this indicated that biguanide is not a n intermediate in the reaction. Copper, which forms an insoluble cyanamide salt that might coat the lime nitrogen and thus hinder extraction, was found to decrease the yield markedly; it forms a biguanide, but also a n insoluble cyanamide salt which may coat the lime nitrogen and thus hinder extraction. UISCUSSION
On the basis of these studies, a process was devised which is represented by Figure 1. Into a 1.36-gallon (Imperial) kettle were introduced 4.86 pounds ammonium nitrate and 0.73 pound of urea. After this mixture was raised to melting temperature of 120" C., the melt was stirred vigorously while 1.09 pounds of lime nitrogen (6570 calcium cyanamide) were added over 20-30 minutes by means of a screw conveyer. Blthough ammonia was evolved copiously, it was dry and hence did not interfere with the feeding. The initial reaction was endothermic, and close control had to be applied t80maintain the 120-125" C. reaction temperature. The ammonia in these experiments was piped to a dry ice trap where it was condensed and weighed. Test with sodium showed that it was almost anhydrous. The melt was maintained for 2 hours at 120-125" C. and was then diluted with 2.7 pounds of water. This hot solution would, in practice, be treated with ammonia and carbon dioxide. Since we used these effluent gases from the process for analytical studies, we neutralized instead wit,h 1.20 pounds of solid ammonium carbonate. The stirred, hot slurry a t 95-100° C. x a s then filtered by suction and the filtrate cooled to 25" C. The hot sludge on the filt,er was washed with 2.75 pounds of fresh water, which was retained as dilution water for the nest cycle. I n plant practice the xashed sludge would be transferred to a gas generator, where dilut,e sulfuric acid n-odd free the carbon dioxide needed for the neutralization stage of the process. (In processes JThere the guanidine nitrate is subsequently to be used for nitroguanidine synthesis, the dilute sulfuric acid is conveniently the spent nitfration liquor.) The original filt'rate, chilled to 25' C., precipitated 1 pound of guanidine nitrate together with about lO%l, of its weight of contaminant ammonium nitrate. This contaminant may be washed out on a centrifuge x i t h 4-870 of the guanidine nitrate. These dissolved salts need not be mhollg lost to the process, since the fresh water (2.7 pounds) may be used for this centrifuge washing after it has been used to vash the sludge. This would reduce the throughput slightly. It would probably be more economical to retain the contaminant ammonium nit,rate, if the product were to be used for manufacture of nitroguanidine. The filtrate from which the product was removed ought to contain 3.91 pounds of ammonium nitrate and 0.73 pound of urea. After analysis for ammonium nitrate (by formalin titration), the water (2.5 pounds) was removed in one pass through a single-tube stainless steel evaporator maintained a t 142' C. by a jacket containing boiling isoamyl acetate. At an internal pressure of 10-20 mm. mercury and an internal temperature of 120" C., t,he rate of evaporation was 0.75 pound of water per square foot of heating surface per hour. This apparatus removed water to leave less than 0.5y0but x-as undoubtedly very inefficient. The residue from which the water was stripped wis conveyed. still molten, back t o the original kettle. The urea content n-as checked by setting point after 0.95 pound of fresh ammonium nitrate (one equivalent plus 0.1 pound lost in product) was added. A small amount of make-up urea was added when necessary, and the cycle was then repeated. A series of thirty consecutive experiments were carried out on the same reaction medium with an average absolute guanidine nitrate yield of 92a/& on which the flow sheet is based. The last ten runs did, however, give a slightly higher (94%) average yield. These yields were based on the analytical isolation of the picrate (25), since the product was contaminated with about 10% of ammonium nitrate. I t was, otherwise, free from dicyandiamide,
Vol. 38, No. 12
melamine, biguanide, or water-insoluble material; but in some runs it contained traces of thiourea, detectable qualitatively by a brown-black precipitate with ammoniacal silver nitrate and quantitatively by iodometric titration. (This titration with 0.01 N iodine should be carried out with a minimum of potassium iodide and a slight acidity contributed by sulfuric acid. The method is restricted to solutions containing not more than 0.0270 thiourea. We prefer this method to either the Volhard silver nitrate or Werner's (26) selenious acid method.) The purity of the guanidine nitrate may be attested by the 95% yield of crude nitroguanidine, melting at 249' C., which could be obtained from it. By contrast, nitroguanidine produced by Welland Chemicals Limited melted a t 236-239" C. x i t h decomposition. Thiourea is formed from cyanamide and the hydrogen sulfide liberated from the calcium sulfide present in the lime nitrogen. I t s formation is therefore a problem in any guanidine nitrate process: HZN-CEN
+NH-C=NH
+NHy-C-NH2
AH
B
Since thiourea is destroyed fairly rapidly in aqueous ammoniacal solution a t 100" c., it never exceeded its saturation value of 5.5 g. per 100 ml. of the solution in the present process. Retention in the final product was mainly due to occlusion but ought to be avoided, since amounts up to 1% tend to cause disagreeable evolution of nitrogen oxides during conversion of guanidine nitrate to nitroguanidine. The destruction of thiourea can be effected by prolonging the holding time of the diluted reaction slurry a t 95-100" C. in the presence of ammonia. This process seems to produce a purer product and to give a higher yield than any other which the authors have examined. I t s disadvantage lies in the large excess of ammonium nitrate required. This disadvantage is not reflected in the size of equipment, which is, over the entire process, more modest than that required by other methods. The consumption of ammonium nitrate (0.95 pound per pound of product) is likewise quite favorable, but the nonexpended excess requires 2.7 pounds of water for its solution to avoid loss of the nitrate and contamination of the product, The single disadvantage lies, therefore, in the evaporation of this water. It was stated earlier in this report that the slow addition of nitrolime enabled the reaction to proceed to high yield a t 120' C., whereas about twice as much excess ammonium nitrate was required for the same yield when the nitrolime was added at once. I n this series of experiments an addition time of more than thirty minutes was not examined; a study of this variable as it affected ammonium nitrate excess %-asleft t o the pilot-plant scale where the related study of water evaporation could be studied more efficiently than was possible in this laboratory. It may be predicted that addition periods up to 90 minutes will decrease thg necessity for such large ammonium nitrate excesses as were used in these studies, Consequently, the operation of water removal, if it should prove onerous, may be simplified. ACKNOWLEDGMEKT
This work was carried out at the request of the Canadian Department of Munitions and Supply. LITERATURE CITED
Bayer & Co., German Patent 267,380 (Aug. 18, 1912). Blair, J. S., and Braham, J. H., IND. END.CHEM.,16, 848-52 (1924). Blair, J. S.,and Braham, J. €I., J. Am. Chem. Soc., 44, 2342-52
(1922). Blair, J. S., and Braham, J. H., U. S. Patent 1,441,206 (Jan. 9, 1923).
Burns, R., Gay, P. F., and Imperial Chemical Industries Ltd., Brit. Patent 507,498 (June 15, 1939).
December, 1946
INDUSTRIAL AND ENGINEERING CHEMISTRY
(6) Davis, T.L.,J. Am. Chem. Sac., 43,2234-8 (1921). (7) Davis, T. L., Org. Syntheses, 7,47 (1927). (8) Davis, T.L., U. S. Patent 1,417,369(May 23,1922). (9) Ibid., 1,440,063(Dec. 26, 1923). (10) Erlenmeyer, E.,Ann., 146,258 (1868). (11) Ewan, I.,and Young, J. H., J. SOC.Chem.Ind., 40,109-21 (1921). (12) Gockel, H.,Angew. Chem., 47,555-6 (1934). (13) Griessbach, R.. and Rossler, A., German Patent 490,876 (Dee. 15, 1925). (14) Hill, W. H a ,Swain, R, C., and Paden, J, H. (toAmerican Cyanamid Co.), U.S. Patent 2,252,400(Aug. 12,1941). (15) Jones, R. M., and Aldred, J. W. H . , IND.ENQ.CHEM.,28, 272-4 (1936). (16) Kat6, Y.,Sugino, K., Koidzumi, K., and Mitsushima, E., J. SOC.Chem. I n d . J a p a n , 36, Suppl. binding 133-4 (1933).
1319
(17) Pinck, L.A., IND. ENG.CHBM.,17,459-60(1925). (18) Sander, F.,Ger. Patent 527,237(Jan. 1, 1928). (19) Schmidt, E.,Arch. Pharm., 254,630 (1916). (20) Smith, G. B., Sabetta, V. J., and Steinbach, 0. F., IND. ENQ. C H ~ M23, . , 1124-9 (1931). (21) Spurlin, H. M. (to Hercules Powder Go.), U. S. Patent 2,109,934 (March 1, 1938). (22) Stickstoffwerke,s., G~~~~~ patent222,252(ace. 30, 1908). H'*Ibid** 600,869(Au& 27 1934). (23) Traube, and (24) Volhard, J., J. prakt. Chem., 9,15 (1874). (25)Vozarick~ *' Chem.f 670 (lgo2). (26) Werner, E.A,, Analyst, 65,268 (1940). (27) Werner, E*A.9 J. C h e m S0C.t 107,715 (1915). (28) Werner, E. A., and Bell, J., Ibid., 117, 1133 (1920).
Influence of Water Vapor on Ozonizer Egciency J
CLARK E. THORP AND WALTER J. ARMSTRONG Armour Research Foundation, Chicago, Ill. Graphs are presented to show the relation of ozonizer efficiency over a large range of absolute and relative humidities. Absolute humidity is shown to influence ozonizer efficiency greatly when it is above 0.001 gram of water per gram of air (dew point, -17' C.), but to have no influence below this optimum point. Relative humidity has no effect on ozonizer efficiency.
S
I N C E 1943 the Armour Research Foundation has been sponsoring a series of investigations (1) leading to the design and construction of industrial ozonizers of improved efficiency. I n this connection the influence of water vapor, temperature, pressure, and frequency on ozonizer efficiency had to be determined. A search of the literature provided a considerable amount of data of academic interest but of little actual engineering value. For example, it is well known that water vapor decreases the energy yield of the reaction 0 2 t o 03 (8) and increases the to O2 (417 but enough data are not energy yield Of the reaction given to allow calculation of the amount of drying required for most economical operation of industrial ozone equipment. An additional reason for further investigating the influence of water vapor on ozonizer equipment is brought about by the recent development of new types of generating elements which produce ozone without the production of large amounts of heat (2, 6). The elements, developed and manufactured by Ozo Ray Process Corporation, are flat, plastic plates with imbedded metallic electrodes. With a life of well over 10,000 hours and a total heat rise of 1' C. per square inch, they are well suited for industrial ozone equipment, although at the present time they are not being used for such equipment. This paper presents the results of an investigation of the effect of water vapor on the efficiency of ozonizers using the new type of generating elements. The general shape of the curves obtained for yield os. absolute humidity should apply t o any type of ozonizer using the silent discharge and adequate cooling.
lute humidity), such as the hair hygrometer, electrical conductivity, etc., were considered either too limited in range or too inaccurate. Theoretical determination of the absolute humidity under known conditions of temperature and pressure presented the most logical solution to the problem of controlling the amount of water vapor passing through the ozonizer. Accordingly air containing water vapor was cooled and brought t o dew point by passage through a series of cold traps a t known temperature. The amount of water vapor then present in the air can be calculated from the standard equation
where H = absolute humidity in grams of water per gram of dry air P = corrected barometric pressure p = vapor pressure of water at a cold trap temperature Below 0'0 c. the vapor pressure of supercooled water is used in preference to the vapor pressure of ice, because even if air is in contact with ice its humidity is more accurately calculated from the vapor pressure of supercooled water (3,5).
.AIR
LINE
K
Figure 1. THEORETICAL CONSIDERATIONS
The determination of the influence of water vapor on ozonizer efficiency requires passage of air containing known amounts of water vapor through the ozonizer, with subsequent analysis of the air stream t o determine the quantity of ozone produced. Simple methods of determining the amount of water vapor in air (abso-
L
J
Flow Sheet for Determining Effect of Water Vapor on Ozonizer Efficiency
A . Air cleaner B. Dew point trap in Dewar C. Humidity-regulating traps in Dewar D. Thermometer, $30' to 1000 c. E. Heat exchanger coil F. Manometer
-
G . Thermometer, +30° to
-1000 e.
FI. Ozonizer
Insulated tank J. Gam absorption bottler, K. Wet test .as meter L. Vacuum pump I.
