1286
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
identical temperature conditions, the degree of dissociation seems to rise t o a maximum (higher K p ) with the propyl alcohols and then decreases with further increase in the molecular weight of the alcohol. The molar heats of reaction of gaseous monomeric formaldehyde with alcohol vapor, calculated from the vapor equilibrium constants by means of the Clapepron equation, have been found t o be about 14,800 calories for each of the alcohol systems examined. A comparison of the aqueous arid alcoholic data shows thar, at, the lower temperatures especially, alcohol-formaldehyde mixtures are less completely dissociated than the methylene glycol derived from aqueous solutions, indicating that the affinity of formaldehyde for alcohol is much greater than for mater. A new hypothesis is presented concerning the distillation phenomena which occur during the distillation of aqueous formaldehyde solutions. The presence of an equilibrium in the vapor phase and the rate of adjustment in the chemical equilibrium for the system methylene glycol-monomeric formaldehyde-kat,e~ seems t o account for many of the phenomena that occur during the distillation of aqueous formaldehyde solut,ions.
Vol. 41, No. 6
LITERATURE CITED
(1) A u e r h a r h , F.,and B a r s c h a i l , H.. A r b . kaiserl. Gesundh.. 27, 7
(1907). (2) ; i u e r b a c h , F., a n d B a i s r h a l l , H., Arb. Reichsgeszindh., 22, 584
(1905). ( 3 ) B o n d , H. A . . U. 9. Patent l,X)5,03:3 ( d p r i l 2 5 . 1 (4) D a n i r l s , F., M a t h e x - s , J . E-I.,arid W i l l i a m s , J. IT., P h y s i c a l Cheiiiiat.r\-," p. 133, Nen; Yolk, McC
Co., 1941. ( 5 ) D e l e p i n e , M . , Compt. ~ r n d .124, , 816,1454. 1528 (1897). (6) H i h b e n , J., J. Am. ('kern. S o c . , 5 3 , 2418 ( 1 9 3 1 ) . (7) Johnson, H G., a n d P i t , e t , E. L.. 1x1). ENG.C ' r % m , ,40, 743 (1!)1S), (8) Ledbury,W., a n d B l a i r . E , , Ilept. Sci. I n d . liesearch (Bi.it,), S p e c . R e p t . 1, 40-51 (1927). (9) L e m n i e , G., Chem. % i o , , 27, 896 (1903). (10) P i r e t . E. L., and H a l l , XI.W,, INI).Ihc:. CHEW,40, 661 (lO4S). (11) S c h o u , 8 . d.,J . ('hrnz. P h y s . , 26, 7% (1929:. (12) $Talker, J. F., " F o i ~ m a l d e i i y d e . " A.C.S. >Ioriograph 98,pp. 1h38, 48-60, i 3 . S e w Yoi,k, R e i n h o l d P u b l i s h i n g Corp., 1944. (13) M'niker, J. F.. .I. A m . Chem. Soc.. 55, 28 (14) Kalker. J. F., J . Plius. C h c m . . 32, 1104 (13) Z i m m e r l i , A , , Inn. 1-h-c;.( ' H E X . , 19, 524 (16) Zinimerli, A., U. S . P a t e n t 1,862,179 (March 13, 1928). KECEIT.E;D .July A, 1948. Inrostigation siipportcd i n part by fund by the G r a d u a t e Scliool. I-nirernity of 3Ijnnesota.
G. ALLEN CAVE, NATHAN J. KROTINGER, AND JOHN D. JIcCALEB Central Experiment Station, U . S . Bureau of Mines, Pittsburgh, P a .
T h e preparation of TNT, PETN, nitroguanidine. and ammonium picrate of various particle sizes and shapes is described. Methods employed include pebble milling, slow cooling with or without stirring, quick cooling in ice, the addition of the hot solvent to dry ice w-ith stirring, and the addition with stirring of the hot solution to a cold diluent. I t is shown that fine particle sizes may be reproducibl) obtained by methods involving the use of dry ice, or by the addition of a hot solution to a cold diluenk.
HE investigation of the effect of particle size upoii the detonat,ion of compressed explosivc charges requires closely sized samples of the materials. This paper describes the preparation of t,rinitrotoluene ( T S T ) . pentaerythritol tetranitrate ( P E T E ) , nitroguanidine, and animoniuni picrate samples of various particle sizes and shapes. There are many rnet,hods for classifying powders in bulk. Of these, conceivably sieving, sedimentation in a liquid column, and gas elutriation were available for use with the niaterials employed. Honxver, sieving, even when effected by hand, n-as ineffective as the materials tended to acquire a space charge and form hard balls in the sieve. Even when t'his effect was not encountered, the classification was inaccurate, depending rather on the second largest cross section of the particle than on the mean diameter. Gas elutriat,ionwas also uiisatisfactory as the part'icles tended to clump, particularly a t the lower size ranges. Sedimentation in a liquid column was, in general, a sa.tisfactoi-ymethod of classification. Hindered settling was minimized by control of the concentration of the solid, and tmbulence mas avoided either by careful temperature control or by increasing the cross section of the settling column. The crystal densit'ies of the explosives ranged around 1.6, and practicable settling times could
be obtained by using saturated solutions of the materials in carbon tetrachloride or mixtures of carbon tetrachloride Jyith methylene iodide, iodobenzene, or n-butyl alcohol. DETERBXINATION OF PAK1'ICLE S I Z E
The particle size of prepared material was determined by microscopic examination or by electron microscope photography. Samples for microscopic examination viere prepared by adding a small portion of the material to a drop of olive oil or castor oil on a microscope slide. Aft,er gentle mixing with a glass rod, the cover slip was added and the slide viewed with the microscope. The length and width of particles in the field were determined by a scaled eyepiece which previous calibration had shown to have a scale division length of 0.5 t o 8.3 ,u according to the magnification used. The depth dimension was determined by focusing on first the top, then the bottom of each particle, the difference in height being shown on a micrometer scale attached to the racking mechanism and calibrated so that the nearest 0.5 p could be estimated. The operation vias repeated using a horizontal traverse for a number of particles in the center of t,wo or three fields. For most samples examined, a count of 200 particles gave scarcely any advantage over a couut of 50. For some wide ranges, hoTvever, counts of 100 t o 200 particles were necessary to give reasonable accuracy. The following parameters were then calculated: 2abc where a, b, and c are t'he longest, second l o n g 1. V,,,, = -, n est, and shortest dimension of each particle, and n is t'he number of uarticles counted. 2.
Mean particle diameter (size) =
4F
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1949 3. 4.
4:
Mean area = -Ratio longest/shortest avis =
Z;alc n
5. The percentages of particles in which the ratio a/c is less than 2.5 and is less than 5.5.
TABLE I. SUMMARY O F METHODS O F POWDER hIa tel ial TNT
A“ Bb Water a n d Benzene l”lp tannic Nitromethane acid CCll
PETN
Water
Acetone n-Amyl acetate E t h y l acetate Ethylene glycol
CC
n-Amyl acetate Benzene
_-~-
PREP.4RATION
ze
~
-~
7
Solvent Diluent CCli Benzene Naphtha Naphtha n-Amyl acetate Xylene Mpthyl alcohol Nitromethane E t h y l ether n-Butyl alcohol Nitromethane CCh Methyl alcohol Water E t h y l alcohol Watei n-Butyl alcohol Water n-Amyl a r e t a t r Sitromethane CCla Benzene Nitromethane Ether n-Amyl acetate CClr Acetone Water Benzene Naphtha l31
Benzene Nitromethane CCla
1287
These parameters were used Nitro... throughout to measure the guanidine size and shape distribution of the particles. The volume n-Amyl acetate E t h v l acetate and area were utilized t o calAininonium culate the mean size and, toWater ... picrate ... E t h y l alcohol gether with the observed range, n-Butyl alcohol Cyclopentanone gave an indication of the Nitromethane (with o r withcontrol of size in any operao u t stirring) tion (see also Table XI). Wet pebble milling. T h e m e a n r a t i o of t h e 6 Slow crystallization. Crystallization b y rapid cooling with ice as external coolant. longest to the shortest axis Pouring hot solution onto d r y ice with stirring; subsequent melting of solidified solvent a n d filtration a t ay Iowa temperature as possible. and the percentage of pare Addition of hot solution slowly and with stirring to a n excess of cold diluent. ticles in which the ratio was less than 2.5 or less than 5.5 were utilized, in the same POWDER PREPARATION way, t o indicate the control of particle shapes achieved in any operation. Small sized samples of powder were prepared by pebble milling C
TABLF; 11
SOLUBILITY OF
~ AT B~~~~~~ ~ poINTOF ~
E SOLVENT
Solrlbility, %------Ni!l?guanidine About 8 Insol. Insol. About 4 About 4 About 2 About 2 About 2 About 2
I _ _ _ _ -
‘olvent Kater Benzene Toluene Nitromethane Cyclopentanone Methyl alcohol E t h y l alcohol n-Butyl alcohol Isopropyl alcohol Ethylene glycol
TNT Slightly Very Very Very Very Very Very Very Very
Acetone Ether Ethyl a r e t a t e n-Amyl acetate CCli
Very Slightly Very Very About 20
PETN Insol. About 5 About 8 About 3 About 1 Slightly Slightly Slightly Slightly .ibout 4 (1000 C.) -4bout 10 Insol. About 3 About 10 Slightly
...
Ammonium Picrate About 16 Slightly Slightly About 2 Very About 6 About 8 About 8 About 10 Very
About 1 Insol. .ibout 2 .ihout 4 Insol.
About 1 Slightly About 2 About 3 Insol.
and by methods involving precipitation from hot solution by cooling in dry ice and by the addition to a n excess of a cold diluent. A general tried is I and~the ~ outline of~ the procedures ~ ~ given in Table , solubility relations are shown in Table 11. The following tables, arranged according t o methods of preparation shown in Table I, give the properties of the powders prepared. TRINITROTOLUENE ( T Y T )
WET PEBBLE MILLING(A). The particle size of powders prepared by this method follow: Alean Particle Size, 1.1 100 g. T N T with 100 g. water and 1000 g. pebbles milled 2 hours: filtered a n d dried a t 60° C. Same. with 0.1 a. tannic acid added before milling
Size Range,
30
90-3 0
24
80-6.0
1.1
111 anot,her effort to obtain finely divided crushed material, 20 grams of cast TNT were melted in a beaker over a water bath. The liquid was then poured onto dry ice and the solid gently Ratio, L ~ ~ % ~Particles / Mean crushed in a mortar. A sample of finely crushed material was then Size Short with Ratio Particle Solution Size, P Range, P Axis in turbulent gas flow by studies of thermal energy transfer. Experiments were conducted in a vertical %inch diameter steel duct with flow velocities ranging from 2 to 7 feet per second. Preliminary results obtained indicate ranges of diffusivitj for certain conditions of turbulence from which qualitative deductions may be made regarding the factors influencing diffusion and mixing. These methods olYer a possibility for a simple and useful tool in the investigation of turbulent fl6w and transfer problems.
E
DDY diffusivit,y or the tendency of a fluid in turbulent flow to disperse material, energy, momentum, or other intensive property is a fundamental factor in several unit operations such as heat transfer, mixing, extraction, and absorption. Diffusivities are particularly basic t o gas absorption, and a more exact knowledge of the nature and magnitude of eddy diffusivity might be utilized to reduce the uncertainties in the application of gas absorption theory t o cases where transfer rates are dependent on both molccular and eddy diff usivities, Therefore, the objective of this investigation was to examine possible methods for measuring eddy diffusivities, t o determine these diffusivities under measured flow conditions, and t o at,tempt to evaluate the factors that influence eddy diffusivity.
G E N E R A L THEORY
11any investigators have explored the analogies between material, moment,uni, and thermal energy t’ransfers Tvithin flowing fluids. Ton-le and Sherwood (12, I S ) , and Sherwooci and TIToertz ( 6 ) have studied diffusion in gas streams and have shown that eddy diffusion is analogous t o molecular diffusion although the eddy diffusivity coefficient is considerably larger. They hypothesized that, an eddy diffusivit,y, De, could be defined for t u r h u l m t Hon- by the equation:
~ v h e Qi ~= rate of transfer of any transferable property per unit area, and dcidy = concent,ration gradient of the property in t h r diffusion direction. Kaliriske and eo-n-orlters (2, 3) found that the distribution of concentrations of injected ~naterialperpendicular to the flow of water in a n opcn channel followed the normal error Ian which is the Fi‘tme type of distribut’ion predicted by diffusion theory. Stutzman ( 7 ) derived a more rigorous equation based on the assumptions fundamental to niolecular diffusion for a p,rocc:ss in which a dye was introduced into the cent,er of a circular pipe in which n-ater was flowing under turbulent conditions. Hi-i experimental data were in agreement Kith the equation and indicated the analogy between molecular and eddy diffusion. Prandt,l (j), Taylor (8-11), arid von Karman ( d ) , have macle fundamental contributions to turbulent flow theory and have advanced the concept. of dividing eddy diffusivity (longth squared per time) into two parameters, intensity of turt)ul(,nce (length per time) and miring length (length).
