1432 FARMER THE VELOCITY OF DECOMPOSITION OF CLVI 11.-The Velocity of Decomposition of High Explosives in a Vacuum. Part I. By ROBERT CROSBIE FARMER. PREVIOUS work on the stability of explosives has been devoted almost entirely to the nitric esters. The methods used depend for the most part on the detection and estimation of the oxides of nitrogen evolved on heating and are not in general applicable to nitro-aromatic compounds or to mercuric fulminate azides etc. I n order to meet the need for a simple quantitative stability test for such compounds the vacuum test described below was developed. This has been widely used for high explosives more particularly for trinitrophenylmethylnitroamine. It has proved itself so simple in use after some thousands of tests that it may be of interest for the investigation of other reactions in which gases are evolved.Some measurements of the rate of decomposi-tion in a vacuum have been made on guncotton (Obermiiller M i t t . B e d . Bezirksuer. 1904 1 30; Dupr6 Ann. Rep. I m p . of Explosives 1903 26; 1904 28'; 1905 29; Hodgkinson and Coote, Chem. News 1905 91 194; Robertson and Napper T. 1907 91, 764; Willcox J . Amer. Chem. Soc. 1907 30 271; Pleus, Zeitsch. yes. Schiess. u. Sprengstofw. 1910 5 121) on silver oxalate (Hoitsema Zeitsch. physilcal. Chem. 1896 21 137) tri-nitrotoluene (Verola Mkm. poud. Salp. 1911-1912 16 40) and tetryl (Knowles J . Tnd. Eng. Chem. 1920 12 246). The methods described were not however convenient as standard tests for high explosives. From measurements on a large number of explosives it appears that these are in all cases subject to a gradual decomposition with evolution of gas a t temperatures below their ignition points.The velocity decreases strongly as the temperature is lowered butl there can be little doubt that a very slow decomposition must occur even a t the ordinary temperature. As in the case of nitric esters (Farmer this vol. p. Sll) the decomposition is partly catalytic and partly non-catalytic ; when catalytic influences are eliminated the velocity sinks to a minimum. I n many cases the catalytic decomposition outweighs the intrinsic decomposition of the pure substance and it is frequently difficult to purify the explosive to such a degree that the catalytic influences are completely removed.For the same reason different prepar-ations of the same substance often differ considerably in their ratas of decomposition Frequently the avolution of gas proceed H-IGH EXPLOSIVES I N A VACUUM. PART I. 1433 with an acceleration due to autocatalysis whilst in other cases it becomes slower after a time owing to the decomposition and con-sequent elimination of impurities. In many cases the presence of moisture gives rise to very erratic results and special steps are therefore taken to eliminate this influence. As a rule the decom-position has been carried only to the extent of a small evolution of gas. The measurement of small volumes of gas has it is true, the disadvantage that the measurements are more affected by traces of volatile matter etc. but if the decomposition is carried further, the products formed are liable to have a very disturbing effect and frequently a very rapid evolution of gas sets in.The nitro-aromatic compounds are in general very much more stable than the nitric ester explosives. Whilst guncotton shows a marked decomposition in Will's test a t 1 3 5 O in four hours and nitroglycerin a t a lower temperature the trinitrobenzene deriv-atives require in general temperatures of 140° to 180° in order to give readily measurable volumes of p s on heating for 100 hours. The dinitro-compounds show scarcely any measurable decomposi-tion. The nitroaminw such as tetryl on the other hand are less stable and decompose sufficiently rapidly for measurements at 1 20°. I n order to give an approximate idea of the relative stability, the temperatures may be calculated by extrapolation a t which the gas evolution amounts to 1 C.C.per gram in 100 hours: Trinitrobenzene .............................. 190-195' Trinitrophenol ................................. 150-158" 2 4 6-Trinitrotoluene ..................... 135-140' 2:3:4- ..................... 135-140" 3:4:6- 11 130-135' Trinitrophenylmethylnitroamine ......... 115-1 20' Cellulose nitrate (N= 13 per cent) ...... Approx. looo 11 ..................... Trinitrotoluenes gave 0.8 to 1.8 C.C. of gas per gram in 100 hours a t 140O. The differences in velocity between different samples indicated the presence of traces of catalysts although most of the samples were very pure. There was practically no parallelism between the melting point and stability.The rate of evolution showed no acceleration ; hence no autocatalysis occurred within limits measured. From the results obtained it is evident that pure trinitrotoluene could be kept indefinitely a t the ordinary temperature but the actual rate of decomposition cannot be ascer-tained by direct extrapolation as the results obtained in the case of tetryl have shown that a great increase in stability occurs on pasging from the molten to the solid condition. The isomeric trinifrotoluenes which accompany the symmetrical compound in small proportion on nitration of toluene were als 1434 FARMER THE VELOCITY OF DECOMPOSITION OF examined. The commercial products contained impurities which decreased their stability. When these were removed by crystal-lisation 2 3 4-trinitrotoluene showed about the same stability as the 2 4 6-isomeride whilst 3 4 6-trinitrotoluene was somewhat less stable.The stability of 2 4 6-trinitrotoluene was not sensibly affected by the addition of 1 per cent. of the other pure isomerides, but 5 per cent. of the 3 4 6-isomeride caused a slight increase of velocity. Mixtures of picric acid and trinitrotoluene showed a lower rate of decomposition at 140° than trinitrotoluene alone. This was somewhat surprising since the general experience is that acids decrease the stability. It is possible that the trinitrotoluene exists as an ec;rnilibrium in which a small quantity of an isonitro-compound is present the latter being the cause of the instability observed.Picric acid miqht readily cause the isonitro-compound to revert to the normal nitro-compound thus increasing the stability. As an example of an unsaturated substance castor oil was mixed with trinitrotoluene and with picric acid and the mixtures were tested. The castor oil depressed the stability very strongly in both cases. Trinitrophenol showed a stabilitv intermediate between that of trinitrobenzene and that of trinitrotoluene. Trinitrobenzene showed extreme stability even a t the boiling point of aniline and the stability was not perceptibly affected by wet and dry storage trials. This is of interest as showing that notwithstanding the difficulty of introducing the third nitro-group the trinitro-compound when once prepared is very stable.EX P E R I M E N T A L . A pparatus.-The thermostat (Fig. 1) consists of a cylindrical copper bath. which is maintained a t the required temperature by a boiling liquid. The cover consists of a thin brass casting with six orifices for the heating tubes and a short column to support the condenser. These all form part of the casting thus avoiding joints which are otherwise very apt to Teak during protracted tests. The connexion with the condenser is made by a conical joint' surmounted by a cap which can be filled with vaselin or other material to lute the joint. The brass column also projects about 0-5 cm. into the interior of the bath and is connected with a cylinder of coarse copper gauze of the same diameter as the tube and about 15 cm. in length. The object of this is to convey the condensed liquid down the centre of the bat,h and ensure unifor HIGH EXPLOSIVES IN A VACUUM.PART I. 1435 heating. This is fitted with a cap and a cup for luting. laggiiig material surrounds the bath. The lid has also1 a small opening for filling the bath. A cylinder of The top of the bath is also F I G . 1.A. -- ,----I I I I I I I 1 'i---;-y I 1436 FARMER THE VELOCITY OF DECOMPOSITION OF covered with a loose cake of asbestos lagging material about 3 cm. thick with holes corresponding with those in the brass cover. The condenser (Fig. 2) is of the multitubular pattern and the water inlet tube is fitted with a funnel as shown so that the flow of water can be seen. For temperatures from 800 to 1000 mixtures of alcohol and water were used.From looo to about 1 3 5 O solutions of calcium chloride were generally taken. In the latter case the addition of a little lime is advisable to avoid corrosion of the copper. The temperature of the boiling calcium chloride solution is FIG. 1 ~ . readily adjusted by adding water if too high or by allowing water to evaporate if too low. The level of the liquid should be main-tained a t about 5 cm. from the lid of the bath. Safety Precautions.-To guard against damage by explosion, steel tubes with closed bottoms were provided; these fitted loosely into the brass orifices and were packed round with fine copper fillings to give good contact with the bath. They were made rather short (Fig. 3) since it was found that if they extended nearly to the surface of the bath they became cooled by radiation, and irregular temperatures were obtained HIGH EXPLOSIVES IN A VACUUM.PART I. 1437 The thermostats were surrounded by screens and placed in a A cistern was provided to guard against failure fire-proof shed. of the water sup-ply and an auto-m a t i c ball-cock tap was fitted to cut off the gas supply in case the cistern b e c a m e empty. Heating Tubes. - T h e g l a s s apparatus passed through numerous modifications i n the course of a large number of m e a s u r e m ents. The earlier pat-terns were fitted with glass taps for exhaustion but it was found much better to avoid these and the s i m p 1 e device shown in Fig. 3 was found to be a g r e a t improve-ment.In order to exhaust the tube a quantity of mercury suffi-cient to fill the upright limb of the capillary tube is placed in the lower cup. This cup is connected with the pump by FIG. 2. means of a rubber stopper. The apparatus is then inclined so that the cup is horizontal and the mercury lies in a pool in the cup, leaving a free passage between t,he pump and the capillary tube 1438 FARMER TITE VELOCITY OF DECOMPOSITION OF After exhaustion the apparatus is returned t o the upright posi-tion and on releasing the vacuum the mercury rises in the F I G . 3. capillary tube which acts as a manometer during the test. The cup a t the top of the' heating tube is luted with mercury, and the apparatus is then ready to be transferred to the thermostat.It is i m p o r t a n t that the stopper be very well ground and lubricated as thinly as possible with a non-reactive lubricant. The thermometer is embedded in sand in a similar heating tube, which is introduced into one of the steel tubes and placed in the bath. Accuracy of the tempera-ture readings is of great importance in view of t h high t e m p e r a t u r e-cosfficients of the decom-positions. Met hod of Working.-Very thorough cleaning of the apparatus is neces-sary as traces of foreign matter have a marked catalytic effect. T h e tubes were cleaned suc-cessively with acetone, benzene and hot chromic acid and were then sub-jected to prolonged wash-ing with water. The manometers were simi-larly cleansed.The corn-plete removal of moisture is also of importance since this frequently gives rise to abnormal accelerations. The explosive is dried a t a temperature well below its decomposition point and weighed quantities are introduced into tho t,est-tubes. These ar HIGH EXPLOSI.VES IN A VACUUM. PART I. 1439 then connected with the capillary tubes the ground stoppers being thinly lubricated. The apparatus is then exhausted to about, ti mm. of mercury by nieans of a Geryk pump and is heated t o a temperature a t which no measurable decomposition will occur (generally SOo) for some hours to remove water. The apparatus is then again exhausted dry air is allowed to enter through a three-way cock attached to the pump and removed onca more by the pump and the apparatus is inserted in the thermostat.The time a t which the heating commences is noted and an arbitrary period is allowed to permit the pressure to settle down before the first reading is taken. This is necessary on account of minute traces of residual volatile matter and in some cases the volatility of the compound itself. In general one and a-half hours have been found sufficient. The height of the mercury column is then read a t intervals in comparison with the baro-metric height. It is convenient to use as barometer a similar tube exhausted and placed near the apparatus. This automatic-ally corrects the reading _for temperature of the mercury and capillary depression. To avoid differences in the level of the mercury in the lower cup it is convenient to fill this up so that it overflows into a dish as the mercury descends in the capillary tube.In taking a number of readings a sliding scale may be used with advantage. The zero1 is set to the mercury level in the baro-meter and the difference between manometer and barometer can then be read off directly. Calculation of Gas Fo1ume.-The calibration of the apparatus includes measurements of the volume of the test-tube the volume of unit length of the capillary and the total length of the three limbs of the capillary tube from the stopper to1 a point on the capillary-tube level with the average height of mercury in the cup. The volume of explosive is deducted from the volume of the heating tube to obtain the net volume. I n calculating repeated readings with the same explosive and Lhe same bath temperature the following shortened method of calculation is useful The bath temperature is practically constant ; the ordinary temperature alters somewhat but as this oiily affects the correction of the volume in the capillary tube it may also be taken as constant (it was in general about 30O).I f the difference between barometer and manometer reading be p mrn. the corrected volume of gas in the heating tube is equal to: A little asbestos wool is packed round the stopper. x y 273 Net gas space x 273 + bath temp. 760 1440 FARMER THE VELOCITY OF DECOMPOSITION OF The corrected volume of gas in the capillary tube is equal to (total 273 p 303 r60 length - 760 +p) x (vol. of 1 mm.) x ~ x = p (total length - 760) x 1 1 (vol.of 1 mm.) x - +pa (vol. of 1 mm.) x -. 843 8 43 Hence the corrected volume is equal to p (a + b) +p2c, net gas space x 373 where (I= 760 (273 +bath temp.) 1 843 b = (length of capillary tube - 760) x (vol. of 1 mm.) 1 843 c = - (vol. of 1 mm.). These constants can be determined for the whole series of measure-ments and the calculation is then very simple. The constants b and c depend only on the calibration of the apparatus; a depends also on the volume of explosive and the bath temperature. Correction for Fluctuations in Bath Temperature.-The fluctuations due to alteration of boiling point with variations of the barometric pressure are of importance as the velocity of decomposition usually increases approximately 100 per cent.for each 5 degrees or 15 per cent. per degree. It is better to apply the correction to the time readings rather than to correct the FIG. 4. Hours. 2 4 6-Tr$nitrotoluene (various samples) at 140'. volume of gas from the individual tubes in the bath. If the deviation of temperature is within about 0*3O it is generally sufficient to calculate a time correction for each day's readings, based on the mean temperature of the bath. Thus a deviation of 0*lo corresponds with a difference of 1.4 per cent. in the velocity, or 0.34 hour per day HIGH EXPLOSIVES IN A VACUUM. PART I. 1441 Measurements of Gas Evolution. 2 4 6-Z1rinitroto1uene.-Measurements at 120° gave very low results (about 0.15 C.C. in 100 hours). A t 180° the evolution was too rapid and gave erratic accelerations which made the measure-ments untrustworthy for comparison.Convenient velocities were obtained at 140° and the fo1lo;wing measurements were made on trinitrotoluenes from different sources (Fig. 4). FIG. 5. t 10 20 30 40 50 60 70 80 90 100 Hours. Trinitrotoluene isomerides at 140". Gas Evolution at 140° (c.c. per gram). Melting point. 20 hours. 40 hours. 60 hours. 81.1" .................. 0.35 0.69 1.04 81.05 .................. 0.39 0.78 1.16 81.05 .................. 0.13 0.33 0.70 81-0 .................. 0.14 0.28 0-54 81.0 .................. 0.16 0-32 0.55 81-0 .................. 0-16 0.35 0.58 81.0 .................. 0-12 0.22 0.37 80.95 ............... s.. 0.25 0-50 0.86 80.95 .................. 0.14 0-28 0.52 110.76 .......... ...... .. 0.33 0.68 0.8ti 80.65 0.26 0.54 0.85 80.65 .................. 0.33 0.7 1 @!)O 78.80 ......,..,. ,,... 0.36 0.71 1-0t) . . . . . . . . . . . . . . . . . . 80 hours. 1.31 1.47 0.94 0.85 0-85 0.78 0.57 1.21 0.83 1.01 1-13 1.017 1-45 100 hours 1.56 1-75 1-13 1-13 1.13 0.99 0.77 1.50 1.10 1-26 1.36 1-24 1.8 1442 FARMER THE VELOCITY O F DECOMPOSITION OF Comparative measurements on dinitrotoluene gave no perceptible decomposition in 100 hours. 2 3 4-Trinitrotoluene.-The following figures show the gas evolution after one and two crystallisations respectively. The purification reduced the rate of decomposition slightly. Cas Evolutioi~ a t 140° (c.c. per yram). 20 40 60 80 100 hours. hours. hours.hours. hours. One crystallisation ... 0.44 0.78 1.07 1.40 1.75 Fig. 5 (4) Two crystallisations ... 0.18 0.46 0.76 1-05 1-36 , (3) Y ? 9 9 0.22 0.37 0.67 1.01 1-38 , ( 2 ) ... 2 4 6-T.N.T. (mean) 0.24 0.40 0.77 1.04 1-29 . ( 1 ) 3 4 6-Trinit~otoluene .-The commercial product was much less stable than the above and decomposed so rapidly a t 140° that the measurement had to be made at 1 2 0 O . The results after one crystallisation are given a t both temperatures. Gas Evolution at 120° (c.c. per g r a m ) . 20 40 60 80 100 hours. hours. hours. hours. hours. - - - Original (sample a) ............... 2.2 3.3 One crystallisation ............... 0.55 1.08 1-58 2.05 2-48 Gas E,Liolutz'on at 140° (c.c. per gram). 20 40 60 80 100 hours. hours. hours.hours. hours. Sample (a) One cryst. 4.7 7-9 - - - Fig. 5 (8) Sample (5) One cryst. 0.89 1-80 2.80 4-10 5-70 , (6) ? 9 9 (7) 9 9 TWO , 0.76 1.38 1.99 2.60 3.23 , ( 5 ) - - TWO , 1.50 3.00 -Even after purification this isa'meride was much less stable than 2 3 $-trinitrotoluene. Mixtures of the Isomeric Trinitrotoluenes.-In order to ascer-tain whether the unsymmetrical isomerides which always accom-pany 2 4 6-trinitrotoluene in small quantity in the crude pro-duct affect the stability the following mixtures were examined HIGII EXPLOSIVES IN A VACUUM. PART I. 1443 Gas Evolution at 140° (c.c. per y a m ) . 2 ~ 3 ~ 4 - 3 ~ 4 ~ 6 - 2"::-T.N.T. T.N.T. Puri- T.N.T., per Sam-cent. ple. - -- a b z, a b b 1 b 1 a 1 h 5 a 5 a 5 b --- a ---fica-tion.. -_ corn]. coml. 2 crysts. coml. 1 cryst. 1 cryst. 1 cryst. 2 cryste. 2 crysts. 2 crysts. 2 cryste. 2 crysts. 2 crysts. Per cent. 100 99 9 8 99 95 95 95 95 99 99 99 95 95 95 20 40 60 80 100 hours. hours. hours. hours. hours. 0-16 0.34 0-57 0.78 0.99 1.03 1.55 2.12 2.60 3.01 0-23 0.80 1.40 2.01 2.53 0.15 0-33 0.65 0.79 0.98 2.23 3.34 4.21 4.99 5.78 0-31 0.77 0.98 1-23 1-52 0-54 0.87 1-10 1-36 1*€9 0.17 0.44 0.52 0.78 0.98 0.48 1.11 1-61 2-10 2.53 0.10 0.67 1.02 1.37 1.62 1.28 2.01 2-85 3.45 4.02 4.60 1-56 2-51 3.22 3.86 4.44 2.7 1 - - - -- - - -Influence of Picric A cicl o n the Stability of 2 4 6-Trinitro-toluene.-A trinitrotoluene was tested alone and in admixture with picric acid.The evolution from the1 mixture was less than from Irinitrotoluene alone. C.C. of gas (corr.) (140"). , Picric 2 4 6- 20 40 GO 80 100 150 200 acid. T.N.T. hours. hours. hours. hours. hours. hours. hours. 2 - 0.14 0.25 0.29 0.35 0.44 0.62 0.83 2 2 0.14 0.22 0.25 0-32 0.42 0-62 0-81 2 2 0.15 0.23 0.28 0.34 0.43 0.68 0.95 - 2 0.25 0.50 0.74 0.96 1.19 1-94 2.88 Iiifluerbce of Unsaturated Compounds.-To avoid a disturbing effect on the gas volume due to vapour pressure castor oil was chosen as a fairly non-volatile unsaturated substance. This increased the rate of decomposition very strongly; even a t 12Q0, the gas evolution from trinitrotoluene containing 5 per cent. of castor oil was readily measurable. C.C. of gas (corr ) (120").1 2 4 6- Castor 20 40 60 80 100 150 200 T.N.T. oil. hours. hours. hours. hours. hours. hours. hours. 5 0.25 0-65 1.00 1.90 2.80 3.75 7.60 11.10 Naphthalene under similar conditions gave no measurable gas evolution in admixture with trinitrotoluene even when present t o the extent of 40 per cent. Picric Acid.-Measurements were made a t 140° and 183" on picric acid crystallised from water. The gas evolution was very rapid a t the higher temperature but showed some decrease in velocity as the decomposition proceeded from which it woul 1444 FARMER THE VELOCITY OF DECOMPOSITION OE' appear that some catalyst was being gradually eliminated. The successive crystallisations did not reduce the rate of decomposi-tion a t 183O' but at 140° some stabilisation was noticeable in the latter crystallisations.Gas Evolution at 140° ( c . c . per gram). 100 hours. 150 hours. 200 hours. 300 hours. Original .................. 0-1 1 0.27 0.44 0.88 Cryst. No. 2 ............ 0.13 0.35 0.51 0-92 , , 4 ............... 0.04 0.12 0.23 0.50 , , 5 ............... 0.06 0.2 1 0.31 0.58 Gas Evolution at 183O ( c . c . per gram). 1 2 3 5 10 15 20 hour. hours. hours. hours. hours. hours. hours. Cryst. No. 1 2-90 4-50 5.45 6.45 - - -) , 2 2.95 4.85 6.10 7.30 - - -, , 3 2-30 3.95 5.20 6.15 8.35 10.60 12.80 ) , 4 1.90 3.80 5.50 6-65 9.00 11-35 13.65 , ,) 5 2.05 3.90 6-65 6.95 9.20 11-50 13-80 As in the case of trinitrotoluene castor oil increased the decom-position. Whereas picric acid alone gave practically no measur-able decomposition a t 120° a mixture of picric acid with 5 per cent.of castor oil gave t.he following evolution : Picric Acid 5 grams. Castor Oil 0.25 gram. 50 hours. 100 hours. 150 hours. 200 hours. C.c at! 120" ............... 0.85 2.50 4.70 6.85 1 3 5-Trinitrobeltzene.-The gas evolution from this compound The following was very low even a t 1 8 3 O (boiling aniline bath). rates of decomposition were obtained with different samples : Gas Evolution at 183O ( c . c . per gram). Mean.. .... 20 hours. 40 hours. 0.05 0.14 0.04 0.09 0.03 0.07 - 0.10 - 0.10 - 0.15 0-04 0.11 60 hours. 0.20 0-13 0.10 0-13 0.13 0.19 0.15 80 hours. 0-27 0.18 0.14 0.15 0.15 0.25 0.19 100 hours. 0.33 0.23 0.19 0.18 0.18 0.30 0.23 I n order to ascertain the effect of storage on the stability, samples were kept for a year a t 50° and tested as above a t 183O HIGH EgPLOSIVES IN A VACUUM.PART I. 1445 20 hours. 40 hours. 60 hours. 80 hours. 100 hours. 0-07 0.13 0.18 0.24 0.30 0.08 0.13 0.17 0.23 0.28 0.06 0.10 0.17 0-22 0.27 0.06 0.07 0.16 0.22 0.26 Mean ...... 0.07 0.11 0.17 0.23 0.28 for a year in a saturated atmosphere a t 4 5 O ' and testing a t 183O. The influence of hydrolysis was examined by keeping samples 20 hours. 40 hours. 60 hours. 80 hours. 100 hours. 0.04 0-08 0.13 0-17 0.17 0-02 0.04 0.12 0.25 0.24 0.01 0.02 0.10 0.11 0.16 Mean.. . . . . 0-02 0.05 0.12 0.18 0.19 The effect of an admixture of picric acid on trinitrobenzene was tried at 150° as the decompmition of picric acid itself is too rapid a t 1 8 3 O . Gas Evolution at 1 5 0 O . a 100 200 300 400 500 hours. hours. hours. hours. hours. Picric acid 0.2 gram Trinitrobenzene 1-8 grams ...... 0.07 0-07 0.20 0-28 0.60 nitrob benzene 1.8 grEtms }.*. 0.07 0.26 0.86 1.45 2-22 Summary. An apparatus is deslcribed for the determination of stability of high explosives by the velocity of evolution of gas on heating in a vacuum. All explosives appear to be liable to a gradual decomposition a t temperatures considerably below their ignition points. The velocity is highly affected by temperature and by the catalytic action of impurities. Trinitrotoluene and the isomerides which accompany it on nitra-tion of toluene do not differ greatly in stability when purified. Trinitrobenzene is much more stable notwithstanding the difficulty with which it is prepared by nitration and trinitrophenol shows an intermediate stability. The t.hanks of the author are due to the Director of Artillery for permission to publish these results. ROYAL ARSENAL WOOLWICH. RESEARCH DEPARTMENT, [Received November 3rd 1920.