1124
T H E ~ J O U R N A LO 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 OXIDATION IN THE MANUFACTURE OF TNTL By ALBERTS. EASTMAN
The nitration of cotton and the nitration of glycerin are carried out a t such low temperatures t h a t they are oxidized only very slightly by the mixture of nitric and sulfuric acids, and the spent acids contain only about I per cent of HNOS04, as determined by titration with potassium permanganate. This substance represents the reduction products of the nitric acid, and is assumed t o be formed by the union of nitrous acid or lower oxides of nitrogen with sulfuric acid in the absence of much water. The nitration of toluene, however, is carried out a t such a high temperature t h a t there is considerable oxidation of the nitrotoluenes with a corresponding decrease in yield of T N T . The spent acid contains about 18 per cent of "OS04 and an equivalent amount of oxygen must either have gone t o oxidize some substance, or else it must have escaped in the free state. The nitrator contains a mixture of the various isomeric I n attempting t o identify di- and trinitrotoluenes. the products of the oxidation of these nitrotoluenes, we isolated 2,4-dinitrobenzoic acid. This is found in part of the charges, but most of them do not contain it in appreciable amounts. The total amount of organic acids formed by oxidation of the methyl group of the toluene has been determined in the spent acid the following way: The nitric acid was removed from the spent acid by evaporation, and the sulfuric acid was removed by adding enough barium hydroxide t o neutralize t h e solution, and filtering off the barium sulfate. The filtrate then contained the soluble barium salts of t h e organic acids, carboxylic acids formed by oxidation of the methyl group of the toluene, and sulfonic acids formed by the action of the strong sulfuric acid in t h e nitration. The barium salts of the organic acids thus obtained were then analyzed by evaporating an aliquot part of the above filtrate t o dryness, and weighing the dry barium salts. Then the residue was ignited in air t o convert the barium salts of t h e carboxylic acids t o barium carbonate and the barium salts of t h e sulfonic acids t o barium sulfate. After weighing the mixture of carbonate and sulfate, the carbonate was removed by extraction with hydrochloric acid and the sulfate weighed. This furnished data from which the loss of foluene due t o sulfonation and the loss due t o oxidation t o carboxylic acids, could be calculated. Assuming one acid group per molecule of toluene, there was 0.39 per cent loss of toluene by sulfonation and 1.24 per cent loss of toluene by oxidation t o organic acid; total loss, 1.63 per cent. The ignition loss from the dry barium organic salts may be regarded as representing the total loss of toluene, and this loss amounted t o 1.86 per cent of the toluene used for nitration, which is a n approximate check. This loss of 1.24 per cent is so small as t o account 1 Read before the Division of 1ndiist.ial Chemists and Chemical Engineers, 57th Meeting of the American Chemical Society, Buffalo, April 7 to 1 1 . 1919.
voi.
11,
NO.
I2
for the formation of only a small part of the 18 per cent "OS04 present in the spent acid, and it was therefore necessary t o look for other oxidation products. The next experiments were designed t o show whether any gases are formed during nitration. For this purpose the nitration was carried out in glass in an atmosphere of carbon dioxide, and the gas evolved was collected over sodium hydroxide solution, which absorbed the carbon dioxide. Separate nitrations were run in air t o determine the amount of carbon dioxide liberated, the dioxide being absorbed in concentrated sodium hydroxide solution. I n these experiments, dinitrotoluene or bi-oil, as it is called, was nitrated to trinitrotoluene. This is the third stage in the nitration of toluene t o T N T . It was found t h a t carbon dioxide, carbon monoxide, nitrogen, and oxygen are evolved during nitration. One hundred parts of bi-oil, containing about I j per cent T N T , liberate 7.39 parts carbon dioxide by weight, 0.495 part carbon monoxide, 0.409 part nitrogen, and 0.016 part free oxygen. The carbon dioxide a n d t h e carbon monoxide together represent the loss of sufficient T N T t o lower the yield b y about 5 per cent. As the yield obtained in the manufacture of T N T is about 86 per cent on the average, this loss amounts t o more than one-third of the total loss sustained in t h e whole process of manufacturing T N T . The composition of t h e gas liberated may vary considerably, even during a single nitration, as is shown by the varying inflammability of the mixture. The proportion of carbon monoxide may be so high under certain conditions t h a t the gas mixture becomes explosive, and this may account for various explosions which have occurred in T N T nitrators, where the top of the nitrator has been blown off, without exploding the charge of T N T . The experimental work described in this paper was carried out,by Dr. W. J. Keith and Dr. J. E. Schott. HERCULES POWDER COXPANY KCNVIL,NEWJERSEY
STUDIES ON THE NITROTOLUENES. 111-BINARY SYSTEMS OF THE COMPONENTS fi-NITROTOLUENE, 1,2,4-DINITROTOLUENE, I ,2 ,4,6-TRINITROTOLUENE1 By JAMESM. BELT.AND CHARLES H. HERTY, Ja. Received July 7 , 1919
I n the nitration of toluene by various methods, some of which are described by Hoffman,2 the product is always composed of several nitrotoluenes. The chief compounds are: p-Nitrotoluene ( M N T ) , onitrotoluene, 1,~,4-dinitrotoluene( D N T ) , and 1,2,4,6trinitrotoluene ( T N T ) . Present also, but in smaller quantity, are vn-nitrotoluene and I, 2,6-dinitrotoluene. For an investigation of the conditions which would give maximum yields of any desired nitrotoluene, it is necessary first t o have a n accurate method of analysis of the product. Analyses by direct 1 This paper is the third of a series dealing with the freezing points and thermal properties of the nitrotoluenes, the investigation having been undertaken at the request of the Division of Chemistry and Chemical Technology of the National Research Council. 2 Bureau of Mines, T e c h n i d Paper 146 (1916).
Dec., :rg19
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
,~.
1125
placed in a larger test tube, serving as an air-jacket, This was then placed in a large beaker of water, the temperature of which was held 4 or 5 degrees below the melting point of the compound. The molten mass was stirred constantly by means of a copper wire stirrer. The temperature fell rapidly until crystals appeared and then rose a little, the slight drop below the melting point being due t o supercooling. The temperature then remained constant for from I O to 2 0 min. This method gives much more accurate values than the small capillary tube method, but requires much larger samples of the substance. MELTING POINT OF DNT-various works of reference give the following temperatures of melting of D N T : 7 o o , l 7 0 . 7 ' , ~ and 7 1 ' ~ ~Bodewig4 gives 7 1 ' , Schiff5 gives 7 0 . 5 ', and Giua6 gives 7 1 '; none of these authors offer evidence regarding the purity of their material. and Kuhlberg' give the melting point 7 0 . 5 ' P U R X F I C A T I O N O F T H E N I T R O T O L U E N E S - C O l ~ ~ e r C ~ ~ Beilstein ~ M N T and D N T were kindly furnished by Dr. C. G. after the substance had been crystallized once from Derick of the National Aniline and Chemical Com- (2%; Millss finds 69.2 '-69.5 ' after the substance had pany, and T N T was obtained from the du Pont Com- been crystallized from several solvents. Taverneo pany. After many recrystallizations from alcohol gives the melting point a t 70' after the material had and benzene and finally from pure alcohol, the three been crystallized several times from alcohol, and Rincompounds were obtained pure, as was evidenced by toul1° finds 69.9-70.2 ' after numerous crystallizations from alcohol. We have found t h a t successive crysthe constant melting points. M E L T I N G P O I N T O F aizNT-The melting point Of M N T tallizations from alcohol yield a product of constant The most probable value has been investigated recently by Rintou1,l who has melting point, 69.5'-69.6'. 69.6' * 0.3') using our determinations and those is also tabulated some of the earlier data. He finds t h a t in which several crystallizations are recorded, viz., repeated crystallizations from absolute alcohol gave those of Mills, Taverne and Rintoul. I n earlier work a product melting a t 51.6'-51.9'. MELTING POINT O F TNT-There is much more accord MillsJ2 using several solvents, obtained the constant in the literature regarding T N T . Smith" adopts the , ~ crystallized melting temperature 5 1.3 '. A u w e r ~ who (80.5' to 80.6")) mean of determinations by Comey12 his makerial three times from alcohol, found the meltby Guia13 (80.65"), and by R i n t o ~ l (80.8' ' ~ t o 80.85'). Other investigators4 ing point t o be 51.5'-52.0'. have used Kahlbaum's preparation with a melting Comey points out the importance of accurate informapoint of 54'; and Beilstein6 quotes Jaworski,6 Bartoli tion on this melting point, in the commercial testing and earlier investigators as authority for the value of T N T . The handbooks have generally accepted the value 82' on the authority of Wilbrand.I6 We have 54'. This value has also been given by Hoffman' in his description of t h e nitration of toluene. A crystallized T N T repeatedly from alcohol and have recent handbooks gives the value 5 2 ', without reference obtained the constant melting temperature 80.3 5 ', It seems probable t h a t this value is correct t o 0 . 3 '. to the authority. MELTIKG POINTS OF B I N A R Y MIXTURES-AS there is It is evident from an examination of the literature t h a t where several crystallizations were made, the no single temperature a t which any given mixture melting point falls below 5 2 ' (51.3' t o 5 1 . 9 ' ) ~but where changes entirely from liquid t o solid, we shall take the no descriptions are given for purification methods the usual meaning of the term "melting point." For value 54' is generally accepted. I n the present work mixtures this is t h e temperature where crystals first we have crystallized many times (3 times or more) appear in the liquid. The accurate determination of and have found for successive crops of crystals the such a temperature is frequently complicated by the constant melting temperature 51.25 '. It is probable phenomenon of supercooling, where the melt may cool several degrees below the normal freezing point bet h a t the true melting point is 5 1 . 5 ' * 0.3'. 1 Smith, "TNT and Other Nitrotoluenes," p 102. For the determination of this melting point, we have 2 Handbook of Chemistry and Physirs, 1919, 157. used the cooling-curve method. Five t o I O g. of the a Kempf, "Tabelle," p 28; Bureau of Mines, Technical Paper 146, p. 7. 4 2. Kryst., 3 (1879), 389, from Jahresber., 1879, 395. pure substance were melted in a test tube, in which was 5 A n n . , 21;s (1884), 264. a thermometer reading t o 0.1'. The tube was then 6 Ber., 47 (1914), 1718;GQZZ. chim. itaZ, [l]45 (1915), 339. methods are not possible, as the percentage of any constituent in D N T is the same as the percentage of t h a t constituent in a mixture of equimolecular quantities of M N T and T X T , and as direct methods would not distinguish between isomers. For the analysis of such mixtures it is therefore necessary t o resort to indirect means. Such methods depend on the construction of a chart or table from measurements of many synthetic mixtures of the components with respect to some physical property or properties. With any unknown mixture it is then possible t o refer its measured physical values t o the chart or table and t o fix its composition. We have used the melting temperatures for this purpose for two-component systems and we shall later publish results upon the melting points of the three-component system of the three compounds M N T , D N T , and T N T .
J SOC.Chem Ind., 34 (19151,60. Phil. M a g . , [4] 60 (1875). 17, from Jahvesber., 1876, 377, J . Chrm. Soc., [3] 41 (1882), from Jahresber , 1882, 103. Z . physik. Chem., 30 (1899), 310. 4 Bartoli, Gazz. chim. ilaZ., 15 (1885), 502; Neubeck, Z physik. Chem., 1 (1887), 657. 6 BeFlstein, Vol. 11, p. 92 Z . Chem., 1865, 220, from Jahresber., 1866, 541. 7 Bureau of Mines, Technical Paper 146 (1916) Handbook of Chemistry and Physics, 7th Ed., 1919, 203. 1 2
* 0
A n n , 166 (18701, 13. Phil. M a g . , [ 4 ] 60 (1875), 17, from Jahresbev , 1876, 377; J Chem. Soc., 4 1 (1882), 27, from Jahresber., 1883, 103. 9 Rec. trav. chzm., 17 (1898), 194. 10 J . SOC.Chem. I n d . , 34 (1915), 6 0 . 11 "TNT and Other Nitrotoluenes," p. 79. 12 THIS JOURNAL, 2 (1910), 107. l a Bev., 47 (1914), 1718; Gaze. chim. ital., (11 46 (1915), 339, 557. 14 J . SOL.Chew. Ind., 34 (1915), 60. 15 A n n . , 128 (1863), 178, quoted by Comey. 7 8
,
1126
T H E J O U R N A L OF 1 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
*
1
0
I
I
I
1
2
3
I
.
4
,
5
6
I
1
7
8
I
1
9 l O
M in u fer FIG.1
fore crystals appear. Under such circumstances a considerable quantity of crystals appear and the composition of the remaining liquid is appreciably different from t h a t of the original melt. The temperature observed in this way is not the same as would be found if the crystallization began regularly, for the composition of the liquid phase in equilibrium with crystals is different in the two cases. For aqueous solutions this error is somewhat adjusted by allowing the mixture t o absorb heat slowly until the crystals just disappear, the last temperature being taken as the “freezing point of the solution.” For such melts as we are here considering, however, the supercooling has sometimes been ignored with corresponding errors in the recorded temperatures of freezing. A common experimental method has been t o study t h e melts by means of cooling curves. Let us examine a cooling curve in order t o determine what point should be selected as the freezing point. I n Fig. I the (uncorrected) temperatures have been plotted during the cooling of a binary mixture of 2 0 per cent T N T and 80 per cent D N T . From A t o B the mixture was moIten and the cooling was rapid. At B, however, crystals appeared, thereby liberating the heat of fusion, and as a considerable quantity of crystals form quickly, the temperature rose t o the point C. From C there is again a cooling which is slower t h a n from A t o B on account of the heat of fusion contributed as the cryst3ls separate. The question now arises regarding the proper temperature to take as the freezing point. I n numerous investigations the supercooling effects have been ignored and the freezing point has been assumed t o be the point a t which crystals first appear. This is the point B of t h e figure. It is apparent t h a t the position of B is dependent on the degree of supercooling, which for
Vol.
11,
No. 1 2
any given mixture may vary for duplicate trials. We have found as much as two degrees variation in t h e temperature of B, depending upon the extent of supercooling. We shall refer t o this interpretation later, in explaining certain differences between our results and those of Giua. A second interpretation of the plot is that the freezing point is the highest temperature attained after crystals have appeared, the point C of the figure. This interpretation is valid for finding the melting point of a pure compound, where there are no changes of composition in the liquid phase, but it is not valid where changes of composition occur. This second interpretation is frequently used in texts on physicochemical methods. The error involved is much less t h a n by the first interpretation, but i t is erroneous as the temperature found is t h e temperature of equilibrium between crystals and liquid, the liquid not having the same composition as the original liquid. I n other words, the freezing point observed applies t o a liquid of different composition from t h a t for which i t is recorded, and the extent of the error depends on the quantity of solid phase and this in turn depends on the degree of supercooling. The third interpretation of the cooling curve makes the point C’ the freezing point of the mixture, obtained graphically by extrapolating the branch D C back t o the branch AB. The intersection of these curves gives the point a t which the solid would appear in case no supercooling occurred, and this is of course the desired point.
SO’]
30’)
loo%MNT
130” wO%TNT
’ FIG.2
The following table contains a comparison of results, using each of the three interpretations of the cooling curve for a mixture of D N T and T N T . I n two tests the degree of supercooling was different, this difference being attributable in part t o differences in temperature of the outside bath, which in one case was 46 O and in the other 4 0 ’. Point B Degrees
Point C Degrees
Point C ’ Degrees
... , . . . . . . , . 49.45 50.0 51.3 ... . . . . . . . . . 48.1 49.0 51.3 It will be observed t h a t duplicate results are given Bath at 46’. Bath at 40’.
T H E JOURN.4L OF [NDUSTRTAL A N D ENGINEERING CHEMISTRY
Dec., 1919
b y taking C‘ as the freezing point, while duplicate results are not obtained by taking either B or C. EXPERIMENTAL PROCEDURE
Five t o I O g. of t h e mixtufe accurately weighed were melted in a test tube provided with an accurate thermometer redding t o 0.1’ and with a copper stirrer. This tube was jacketed by a larger tube, which was placed in a 1500 cc. beaker of water, t h e temperature of which was held 4-5’ below t h e freezing point of the mixture. At first the temperature fell rapidly until crystals appeared, then rose a little and again fell, more slowly however than t h e original liquid. By the graphical method the temperature of freezing was found. Duplicate trials gave results agreeing within 0.2 ’. After the primary freezing point had been determined, t h e mixture was further cooled until the eutectic temperature was reached. After slight supercooling t h e second solid phase appeared and the temperature remained constant for a period of time, depending upon the proportion of the liquid which remained t o solidify. I n each binary system the eutectic temperature was constant so far as certain experimental difficulties permitted. When t h e mixture was of high percentage in either component the quantity of solid was so great when the eutectic temperature was reached t h a t t h e mixture could not be stirred. It is obvious that this h a y result in irregularities if in one portion of the paste the stable solid has appeared and in another portion i t has not yet appeared. The lack of stirring may cause also a lack of uniformity of temperature in the mass. When the original mixture was nearer t h e eutectic mixture in composition these difficulties were not met and satisfactory concordant results were obtained for t h e eutectic point.
1127
BINARY SYSTEM: M N T , TNT-Table 1 and Fig. 2 show the freezing points of a number of mixtures of these components. There are two branches t o the curve, one where t h e solid phase is M N T and the other where the solid phase is T N T . These curves intersect a t the eutectic point found for these components.
1.400 0% DNT
iOO% TNT
FIG.3 DISCUSSION O F RESULTS
These binary systems have been studied also by Giual who has given somewhat different data. Our results are in close accord with those of Giua for t h e melting points of t h e pure components, except for M N T where we find 51.25’ and Giua gives 53’ and 54’ in the two publications cited. We also agree on t h e eutectic points. The position of the freezing-point curves is lower according t o Giua t h a n according t o our measurements. This difference we attribute t o TABLEI-BINARY SYSTEM: ~-NITROTOLWENE-TRINITROTOLWENE Giua’s interpretation of cooling curves, for he gives the PER CENT BY WEIGHTMOLECWLAR PERCENT Freezing Eutectic Solid freezing point as “Beginn der Krystallization” and MMT TNT MNT TNT Point Point Phase 100 0 100 0 51.25’ ..... “initio della cristallizzazione.” It seems highly prob90 10 93.71 6.29 48.0 . . .., 80 20 86.88 13.12 able t h a t Giua took t h e point B of Fig. I and therefore 70 30 79.44 20.56 40.1 33.85 60 40 71.31 28.69 35.1 33.95 read temperatures lower than t h e true freezing points. 55 50 40 30 20 10
45 50 60 70 80 90 100
66.93 62.35 52.47 41.52 29.29 15.54 0
33.07 37.65 47.53 58.48 70.71 84.46 100
36.35 41.35 50.0 57.9 65.75 73.1 80.35
34.0 34.2 34.25 33.7
0
69.54’ 65.7
.....
55.65 50.35 50.4 54.3 61.5 68.5 74.4 80.35
45.4 45.6 45.55 45.55
.....
I
TNT
..... .....
0 TAl3LE 11-BINARY SYSTEM: DINITROTOLWENE-TRINITROTOLUENE PER CENT BY WEIGHTMOLECULAR PER CENT Freezing Eutectic Solid DNT TNT DNT TNT Point Point Phase 100 90 80 70 60 45 40 30 20 10
0 10 20 30 40 55 60 70 80 90 100
0
100 91.82 83.28 74.42 65.16 50.51 45.39 34.82 23.76 12.17 0
8.18 16.72 25.58 34.84 49.49 54.61 65.18 76.24 87.83 100
/
.....
.....I 1 ..... .....
TNT
.....
TABLE 111-BINARY SYSTEX: ~-NITROTOLUENE-DWITROTOLUENE PER CENT MNT 100 90 80
70 60 50 45 40 30 20 10 0
WEIGHT MOLECULAR PER CENT DNT MNT DNT 0 100 0
BY
10 20 30 40 50 55 60 70 80
90 100
92.28 84.16 75.61 66.59 57.05 52.09 46.97 36.28 24.93 12.86 0
7.72 15.84 24.39 33.41 42.95 47.91 53.03 63.72 75.07 97.14 100
Freezing Eutectic Point Point 51.250 47.35 36.95 31.1 28.65 33.45 37.8 47.3 55.41 62.45 69.54
.....
Solid Phase
..... 26.54 26.48 26.54 26.44 26.48
..... .....
..... .....
I
c
DNT
FIG. 4 1
Ber., 47 (1914), 1718; Cam. chim. ital., 111 45 (1915), 339.
T H E J O U R h T A L OF I N D U S T R I A L A N D ENGINEERING C N E M I S T R Y
1128
I n many of our cooling curves we find t h a t the temperature of point B , Fig. J , a t which crystals first appear, agrees with t h e values recorded by Giua. We have plotted his results on Fig. 3 for DNT-TNT. These freezing-point curves do not intersect a t t h e eutectic point, a condition which is interpreted as indicating the existence of a compound of t h e components. Of course, if Giua’s data are taken as correct, this must be t h e proper interpretation of the curves.
80“-
Vol.
11,
No.
12
and it is concluded t h a t the mixtures obey t h e ideal solution law. The curves for D N T do not lie so close together. The effect of p-toluidine on the freezing point of D N T is from t h e results of Mremann and Petritschek-* The cause or causes of the deviation may be found when anoQer investigation now in progress has been completed. The formula based on “ideal solutions” involves the latent heat of fusion, regarding which no data could be found in the literature. The direct determination of t h e latent heat of fusion of all of these nitrotoluenes will form another paper of this series. SUMMARY
I n this paper we have given a discussion of t h e melting points of M N T , D N T and T N T , a discusqion of t h e interpretation of cooling curves, and t h e d a t a for the three binary systems of these nitrotoluenes.
70’-
UNIVERSITY O F NORTH CAROLINA CHAPEL HILL, N. C.
60’-
tb
STUDIES ON THE NITROTOLUENES. IV-THE THREECOMPONENT SYSTEM: P-NITROTOLUENE, 1,2,4DINITROTOLUENE, I ,2,4,6-TRINITROTOLUENE2
u, h
By JAMESM. BELLAND CHARLES H. HERTY,JR. Received July 17, 1919
$ 50?
ki
40%
30“
20’ 100
Molecu/ar Per cent FIG.5
Thus in all three of these binary systems Giua claims the existence of compounds over a narrow range of concentration. I n none of t h e cases did we obtain a result other t h a n t h e intersection of t h e freezing-point curves a t t h e temperature found for the eutectic point. Giua’s conclusions therefore regarding t h e existence of molecular compounds in these three cases seem t o be erroneous. In t h e first paper of this series,’ t h e effect of various nitrotoluenes on t h e freezing point of trinitroxylene was studied and the results were found t o accord fairly well with a formula based on the assumption t h a t these mixtures were “ideal solutions.” It is possible t o test the validity of the assumption of “ideal solutions” in another way. Thus t h e freezing point of such a mixture should depend on the molecular fraction of t h e freezing component and not on the nature of the other component or components, provided the components are of like chemical character. When, therefore, the freezing-point curves are drawn using molecular percentages, the curves should coincide. I n Pig. 5 are shown t h e freezing points of t h e nitrotoluenes as influenced by other substances. The curves for M N T and t h e curves for T N T fall quite close together, 1
THISJOURNAL, 11 (1919), 1025.
I n t h e third paper of this series we have given t h e results of a study of the three binary systems of t h e nitrotoluenes: $-Nitrotoluene ( M N T ) , I , 2,4-dinitrotoluene ( D K T ) , and I , 2,4,6-trinitrotoluene ( T N T ) . The present paper contains the resuIts of a study of t h e ternary system of these components. We have again employed the cooling-curve method of finding the temperatures a t which t h e various solid phases appear, using t h e extrapolation method of overcoming t h e difficulty introduced b y t h e phenomenon of supercooling. This method was described in detail in t h e paper above referred to.3
Time
L
Y
U
Z
FIG.1
For a binary mixture there are two freezing points: The primary freezing point a t which t h e first solid appears, and t h e binary eutectic point where t h e two solids are in equilibrium with t h e melt. For a ternary mixture there are three freezing points: The primary freezing point a t which t h e first solid appears, t h e secondary freezing point a t which the second solid appears, and the ternary eutectic point where all three solids are in equilibrium with t h e melt or eutectic mixture. The last temperature was not difficult t o determine 1
Monatsh., 88 (1917). 385.
* This
paper is the fourth of a series dealing with the freezing points and thermal properties of the nitrotoluenes, the investigation having been undertaken at the request of the Division of Chemistry and Chemical Technology of the National Research Council. 8 THIS JOURNAL, 11 (1919). 1025.
