1700
INDUSTRIAL AND ENGINEERING CHEMISTRY
TABLE V.
GELATIONOF ALLYLSUCR~SE IN THE PRESEKCE OF ACRYLIC ASD METHACRYLIC ESTERS
Allslsucrose
+ equal amount of
Methyl methacrklate m-Tolyl acrylate Butyl methacrylate Phenyl methacrylate 2-Ethyl hexyl acrylate Octyl methacrylate Isopropoxy ethyl acrylate Cyclohexyl acrylate Tetrahydrofurfur51 methacrylate
Gelation Time
Min. 210 107 135 1OG 87 69 51 38 37 19
The gelation time for allglbutylsucrose was 300 minutes, and Cor allylbenzylsucrose, 590 minutes. Both compounds gelled considerably more slowly than allylsucrose. Methacrylyl ester of allylsucrose was prepared by dissolving incompletely allylated suciose in pyridine and treating it with methacrylic anhydride. The resulting yellow viscous liquid ( 7 ~ 2 6 1.4788) coiitained 6 7 allyl groups and 1.2 methacrylyl groups. I t gelled in 78 minutes. The original allylsucrose gelled in 205 minutes. The mixture of equal parts of the original allylsucrose and this allylmethacrylylsucrose gelled in 146 minutes, the average time for the two products. Since methacrylyl groups added t o the allylsucrose molecule shorten cor$derably the gelation time of the product, it seemed appropriate to determine whether the mere addition of esters of acrylic and methacrylic acid will not produce the same result, perhaps through copolymerization of the two compounds. T h e results are summarized in Table T'.
Vol. 41, No. 8
Methyl methacrylate had little effect in reducing the gelation time of allylsucrose because the gelation experiments were carried out at 10O0C.,and most of the ester, which boils a t loo", evaporated from the mixture. When the esters of higher boiling point were used, the gelation time was reduced considerably. ACKhOW LEDGMENT
The authors are grateful to E, 13. Bogen for experimental assistance in this work; t o E. 11.Filachione and C. E. Itehberg for providing some of the acrylic and methacrylic esters; to H. W. Strover for preparing the coated panels, to P. E. bleiss for operating the Weather-ometer, and to both for help in evaluating the samples tested. LITERATURE CITED
Frey, Karl, Dismtation, Eidgenos. Tech. Hochschule. Z u r i c h , 1926.
Juvala, Arvo. B e y . , 63,1989 (1930). Nichols, P. L., J r . , Wrigley. A . N., and Yanovsky. E!ias, J . Am. Chem. SOC.,68, 2020 (1946). Nichols, P. L., Jr., and Yanovsky, Elias, Ibid., 67, 46 (1945). Nichols, P. L., Jr., and Yanovsky, Elias, Suaar, 42, KO,9, 28 (1947). O g g , C. L., Porter, W.L., and Willits, C. O . , IND. ENG.CHEM., ANAL.ED.,17, 394 (1945). Salaberg, P. L., U. S. Patent 2,037,740 (1936). Tomecko, C. G., with Adams, Roger, J . Am. Chem. SOC.,45, 2698 (1923). Wrigley, A . N., and Yanovsky, Elias, I b i d . , 70, 2194 (1948). RECEIVED June 30, 1948. Presented before the Division of Sugar Chemistry and Technology a t the 112th Meeting of the ANERICAS C H E w c b L S o i I m y , New York, N. Y.
Thermal Decompositio Nitroxvlene and Nitrobenzene d
P. C. CONDIT AND R . L. HAYNOR, CaliJornia Research Corporation, Richmond, CaZiJ.
S
EVERAL years ago an investigation was undertaken in these laboratories to synthesize aromatic amines. The method chosen involved the catalytic hydrogenation of aromatic nitro compounds a t temperatures of 250" C. and above (4). Nitro aromatics have seldom been handled under such severe conditions, and it seemed possible t h a t euplosion hazards might be involved. A study was therefore made of the stability of these compounds under conditions which might be expected in a commercial hydrogenation. Berthelot ( 1 ) described the behavior of various nitro compounds under thermal shock in 1887. H e explored this behavior by dropping small quantities of t h e compound into a test tube, heated t o a predetermined temperature and maintained in a n inert atmosphere. H e demonstrated t h a t practically all aromatic nitro compounds would explode. The exact temperatures a t which they detonated depended not only on the compound itself but on the size of the sample, the condition of the glass surface, and other variables. I n 1919 D a t t a and Chatterjee ( 5 )reported the effect of structure on the decomposition temperatures of aronlatic nitro compounds. They used Berthelot's methods and showed t h a t the decomposition temperatures mere lowered by increasing the number of nitro groups. Their d a t a appear t o be consistent among themselves, b u t because of the conditions employed the data have little practical value for our purpose.
Bro'ivn, Smith, and Scharmann (5') recently gave data on the thermal decomposition of nitroxylene. I n general, inany of their observations appear to agree with the present authors'. They gave no details on their method of rneasui ement, however, nor any data on the effects of process variables on the decomposition temperatures. Two nitro compounds were used in this investigation, nitrobenzene and nitroxylene. The latter appeared to be the more critical, and it was therefore used in most of the work t o be discussed. ANALYTICAL METHODS AND MATERIALS
Several nitro compounds were analyzed by the method of Kolthoff and Furman (6) with titanous chloride as reducing agent for the nitro group. Results were calculated as weight per cent of nitro groups and were accurate t o =tO.5%. Nitrobenzene was obtained by the distillation of a sample of commercial material. The fraction taken boiled at 210-211 C. and had the following physical characteristics: n % O = 1.5523, dgg = 1.2051. Nitroxylene was also obtained by the distillation of a commercial sample made from mixed xylenes of petroleum origin. The fraction employed had a boiling range of 79 O to 92' C. at 2 mm. It had a refractive index of 1.5443 a t 20" C. and a s ecific gravity of 1.030 at 23 C. Analyses showed 30.0 t o 3 0 . 4 4 of nitro groups (theoretical, 30.4). For a number of runs, samples of impure or semipure nitroxylene from a commercial source were used, analyzing as follows: A, 31.5% NO*; B, 32.2%; C, 32.7%.
August 1949
INDUSTRIAL AND ENGINEERING CHEMISTRY
4,6-Dinitro-m-xylene was employed in a few runs as a typical dinitro compound. It was made from Eastman Kodak m-xylene by the method of Borsche ( 2 ) and was recrystallized twice from a large volume of ethanol. A sample of hydrogenation catalyst was used in some of the work to be described. Directions for a typical catalyst preparation follow: 61.5 grams of commercial 10-14 mesh, granulated activated charcoal were soaked in a solution containing 66 ml. of concentrated aqueous ammonia, 44 ml. of water, and 23.5 grams of molybdic acid. The excess solution was drained off, and gaseous hydrogen sulfide passed through the mass until it was thoroughly saturated. It was then blown with nitrogen and superficially dried in a n open dish on the steam plate. Analysis showed it t o contain 7.23% molybdenum. Before use the catalyst was heated to 260" t o 315" C. in a slow stream of hydrogen. The remaining water and ammonium sulfide were expelled during this operation. 811 other chemicals were either laboratory or good commercial materials.
1701
SECTION A-d
(NOT
SHOWN BELOW)
WITH SUPPORTED BOAT (SHOrm BELOW)
LONGITUDINAL CROSS-SECTION OF AUTOCLAVE WITH SUPPORTED BOAT Figure 1. Arrangement of Samples for Thermal Decomposition Studies
APPARATUS AND PROCEDURE
*
A conventional 2.5-liter chrome-vanadium steel Aminco hydroa small amount of dark-colored oil in the bomb itself, outside of genation bomb was chosen for these measurements. The thermthe boat. Too little of the oil was available for analysis or couple for measuring decomposition temperatures was contained identification except in a few of the nitrobenzene runs to be in the usual well which projected from the head of the vessel into described. the gas space near the sample. A second thermocouple between the heating jacket and the bomb was used t o control the heating RESULTS WITH NITROXYLENE rate and avoid excessive wall temperatures. The bomb was fixed in a horizontal position. In view of the literature, it seemed desirable first t o deterTwo different hemicylindrical glass boats held the samples. mine the effects of sample size and type of equipment on the The first was 9 cm. long and about 5 cm. in diameter, and was ground t o fit the inside diameter of the bomb closely. The second decomposition temperature of nitroxylene, Table I gives the Was 9 cm. 10% and 3 cm. in diameter, and Was supported on four Most of the were made on the purified nitroxylene, radial glass legs, 1 cm. in length. Each boat had a small glass but two runs on commercial sample A were included for comparito one end to facilitate handling, pigure 1 shows the loop son. general arrangement of the sample boats in the autoclave. I n making a run the bomb was wiped out and the boat cleaned Run 79 shows that a 2.1-gram sample of nitroxylene failed to with chromic acid. The sample was weighed into the boat, and entirely but seemed t o d p l l out of the boat and react the boat slid into the bomb until it was about 15 cm. from the or decomPose in the vapor space or on the waI1s of the bomb. closed end. The head of the bomb.was then bolted in place, and I n the range 15 t o 20 grams the decomposition temperatures were the interior purged several times with hydrogen. The bomb was filled with hydrogen t o a predetermined pressure and heated as independent of sample size. Runs on larger samples might rapidly as possible within 10 to 15 O C. of the anticipated decompohave been desirable but uIere not made because of the obvious sition temperature. Heating was then reduced t o a point where hazard involved. Turenty grams were chosen as a convenient the risein temperature wvas 1Oto 2°C. per minute, and temperature and safe sample size for further work. and pressure readings were taken every minute. The slow heating rate during this period was desirable t o compensate for the Comparison of runs 85, 80, and 82 with 90, 93, and 116 intendency of the thermocouple in the well t o 1% behind the wall dicates that the decomposition temperature of both pure and temperature of the bomb, Figure 2 gives temperature and prescrude nitroxylenes as measured in the close-fitting boat were some sure curves for a typical decompo~itionrun on & 20-gram sample 20" C. higher than those measured in the boat supported on legs. of nitroxylene. The point a t which the sample exploded is indicated by a sharp rise in temRun 166 was made with no perature and pressure, and boat a t all and checked can be located accurately the results on the closefrom such a curve. It was found from a number of runs T h e thermal decomposition of aromatic nitro comfitting boat. I n either of to be reproducible for any the latter two cases the pounds has been studied under conditions approximating given set of conditions withsample was in heat-transfer those of commercial hydrogenation. When heated rapin ~ 3 C.' relationship with the heavy idly under pressure, mononitro aromatics decompose with After a run was complete, metal walls of the bomb, explosive violence. A method has been developed for the bomb was cooled and the products were examined. and it is probable that heat measuring the minimum decomposition temperatures reWhen a violent decomposigenerated as incipient deproducibly. Nitrobenzene decomposes at about 356" C. tion took place, the origicomposition set in was carried and nitroxylene at about 308" C. The influence of a numnal nitro compound was aloff comparatively rapidly. ber of different variables on these decomposition temperaways converted t o a black, fluffy mass containing sparI n the supported boat this tures has been investigated. When the nitro compounds kling crystalline specks reheat could be carried away are heated with small quantities of the corresponding sembling graphite. This solid by gas convection only and amines, the decomposition temperatures are lowered. material was in and around so had a n opportunity to When larger concentrations of the amines are present, the sample boat and usually build up in the liquid samhad an ammoniacal odor. the mixtures no longer explode, but relatively large quanElementary analysis showed ple. I n a commercial hytities of dense, resinous solids are formed. If nitroxylene it t o be predominantly cardrogenation, conditions alone is heated for protracted periods at temperatures bon with varying amounts would probably approach below its decomposition temperature, a large part of it is of nitrogen, oxygen, and hythose of the close-fitting converted to a dark resin. Nitrobenzene is only slightly drogen. I n addition there affected by such treatment. was some water and usually boat. It was felt, however,
INDUSTRIAL AND ENGINEERING CHEMISTRY
1702
O F APPARATUS AND SAMPLE SIZE O X THERMAL TABLE I. EFFECT
BEHAVIOR OF NITROXYLENE
Run So.
Grams of Sample
Typeof h-itroxylene
Decompn. Temp.,
Type of Boat
c.
79
Pure
2.1
85
Pure
15.0
Same
331
Pure Commercial A Pure Pure Commercial .I Pure
20.1 15.2 15.1 19.9 19.9 20.2
Game Same Supported Same Same None
331 321 310 309 301 332
80 82 90 93 116 166
TABLE
Close-fitting
None
Appearance of Product Water a n d oil in bomb Light, fluffy carbon Same Same Same Same Same Same
11. EFFECTO F PRESSURL O S THERMAL BEIIA\'IOR PCREXITROXYLESE Run SO.
158 155 93
Pressure, Lb./Sq. Inch Gage Initial Decompn. 250
,575
600
1000 3030
1460
D Temp.,
OF
~ C
~
304 309 309
TABLE 111. EFFECT OF HEATISGBELOW DECOJIPOSITIOS TEMPERATCRE o s THERMAL BEHAVIOR OF KITROXYLESE Run
KO,
119 123 126 120
Type of Sample Pure nitroxylene Same Same Product from r u n 119
Max. Hr. a t Temp., AIax. ' C. Temp. 282 2. 3 254 2.5 2 8 7 % 0 334 0
A s w a r a n c e of Product Dark, i,esinoua solid D a r k liquid Same
Dark, resinous solid
ILL. /O
.,.
22.4 2.2...4
Vol. 41, No. 8
tive, they are of help in interpreting the behavior of the samples on hydrogenation ( 4 ) . Commercial nitroxylene often contains varying amounts of polynitro compounds. One object of this study was to obtain information on the effect of these impurities on the thermal stability of the mononitroaromatic. Runs 98, 101. and 104 (Table IV) wei e made on synthetic samples prepared by dissolving 4,6dinitro-In-xyiene in pure mononitroxylene. The dinitro compound lowered the decomposition temperature sl~ghtlg. I n Figuie 3 the decomposition temperatures are plotted against dinitrovylene content. Runs 106, 117, and 116 (Table IV) were made on the three impure commercial nitroxylenes. The dinitroxylene contents were calcalated froin the nitro group analysis on the assumption that only mono and dinitro compounds \?-erepresent. The data points from these runs are close to the previously established curve ~ ~ ~ ~ ~ , of Figure 3. It seemed possible that the hazards of handling nitro aromatics might be mitigated by blending with aromatic amines. At some point during a hydrogenation these two materials would have t o exist concurrently. Therefore the effects of aromatic amines were determined on t'he decomposition of the corresponding nitro compounds (Table 1%'). Contrary to expectaLions, the presence of xylidine actually lowered the decomposition temperature of nitroxylene, a t least up to 50% dilution. When 25 % xylidine was present, the decomposition product was the usual fluffy carbon. At higher xylidine concentrations the product was a resinous mass, and the violence of t,he decomposition was marlcedly reduced. There was no question, however, t h a t the aromatic nitro compound had react,ed with the amine a t the higher dilutions. The react'ion products were too intractable to identify, but it seems probable t h a t this was an oxidationreduction or a condensation reaction \vhich liberated heat and so caused the nitro compound t o explode a t a lower ambient temperature than normal. I n all but one of the runs reported so far the sample of nitro compound was in contact only with a hydrogen atmosphere and a clean glass surface. This is a condition which could never exist in a commercial operation. The runs of Table V illustrate the effects of varying the gas surrounding the salnple and the surfaces with which i t is in cont,act. Runs 95 and 147 demonstrate t h a t the decomposition tempera'ture of nitroxylene is t h e same in either a hydrogen or a methane atmosphere, which indicates that hydrogen played no part in the thermal decompositions described so far. R u n 151 shows t h a t stainless steel has no effect on the decomposition temperature. The same point was established indirectly for chrome-vanadium steel in run 16s (Table I) where no salllple boat 'ivasused, into intimate The nitro aromatic would also with the cat,alyst in commercial hydrogeuation. R~~ 131 was made in the presence of 10 cc. of molybdenurn sulfide-active carbon catalyst. The decomposition temperature was some 59" C. lower than normal. T h e appearance of the product was
t h a t for test purposes an effort should be made to d e t e m i n e the lowest possible decomposition temperatures, and the supported boat was chosen for further w o r k Table I1 gives data on three runs designed to show the effect of varying the applied hydrogen pressure. The runs were made \pith 2o-gram samples of pure nitroxylene and demonstrate that,, in the range 575 t o 3030 pounds per square in& gage, the position temperatures are substantially independent of pressure, F~~most of the further work the bomb was charged with sufficient hydrogen to give a pressure of about 200 atmospheres a t the anticipated decomposition temperature. It seemed likely that t8herate of heating would have an effect on the decomposition temperatures of aromatic nitro compounds. With the heavy equipment necessary for this work, it was difficult t o vary the heating rate over wide limits and still maintain the degree of control required. .49 a substitute for direct measurement of the effect of heating rate, several runs were made in which samples were raised t o temperatures well below the known decomposition points, maintained a t these temperatures for varying periods, and examined after cooling (Table 111). I n runs 119 and 123 the samples were markedly changed in appearance. T h e dark liquid of run 123 was anaT A B L E IL'. EFFECTO F DINITROXYLENE AND XYLIDISE 0 3 THERMAL BEHAVIOR lyzed by the titanous chloride method OF NITROXYLEKE and ?vas found t o have lost 24.9% DinitroDecompn. of its nitro groups. I n run 126 an Run xylene, Xylidine, Temp., 'O' Type of Nitroxylene 92 % c. Appearance of Product almost identical effect was produced 93 Pure 0 0 309 Light, fluffy carbon by heating the sample to 287" C. 98 Pure, + 4,6-dinitro-m-xylene 5 0 303 Same 10 0 301 Same and allowing it to cool as rapidly as 15 0 299 Same 106 Commerriai Commercial C 1 3 , 7a 0 302 Same possible. Run 120 demonstrates that 117 1 0 .7 a 0 301 Same once a sample undergoes this slow de116 Commercial 4 6.4Q 0 301 Same 113 Pure, + xylidine 299 Same 0 25 composition, it no longer explodes even 167 Same 0 60 Sone D a r k , resinous aolid 108 Commercial C + xylidine 10.30 25 292 Light, fluffy carbon if heated well above its normal decorn111 Same 6.8= 50 302 Heavy, sticky carbon, eome oil Calculated froin analysis for NO2 groups. position temperature. Although the conclusions on these runs are qualita-
INDUSTRIAL AND ENGINEERING CHEMISTRY
August 1949
1703
RESULTS ON NITROBENZENE
Table VI gives the results of a brief investigation of the thermal behavior of nitrobenzene. Run 201 indicates t h a t it explodes under the "standard conditions" a t 356" C., some 48"higher than does nitroxylene. The decomposition seemed t o be somewhat less violent than that of nitroxylene, and the temperature measurement was probably not so accurate. The product was indistinguishable from that previously obtained and was entirely in or around the boat.
TABLE VI. THERVAL BEHAVIOR OF SITROBENZESE Run No.
TIME -MINUTES Thermal Decomposition of Pure Nitroxylene
Figure 2.
typical of the materials obtained from other runs where an explosion had taken place. Runs 137 and 143 were made t o determine the reason for this behavior. The fist demonstrated that molybdenum sulfide was essential t o the lowering of the decomposition temperature observed in run 131. R u n 143 was therefore made in a methane atmosphere. The decomposition temperature was normal. When both hydrogen and molybdenum sulfide are present, the decomposition temperature is significantly lowered. This could mean only that hydrogenation of the nitro compound set i n a t about 249' C., and t h a t this exothermic reaction raised t h e temperature of t h e sample sufficiently above the ambient temperature t o initiate t h e usual explosion. The case is analogous to t h a t postulated for mixtures of the nitro compounds with amines, but the interpretation is clearer in this case since t h e initiating reaction can be formulated with certainty.
TABLEV. Run NO.
EFFECTOF FOREIGN SURFACES ON THERMAL BBHAVIOROF PURENITROXYLENE
Atm.
Foreign Material None Il1lg:Y
137 143
HS CHI
>denurnsulfide on active C Active C Molybdenum sulfide on active C
Quantity, cc.
Deoomp. Temp., C.
10
249
10 10
306 307
*.
Sample
Treatment
Decompn. Tpp.,
C.
201
Nitrobenzene
Regular procedure
388
202
Nitrobenzene Nitrobenzene
205
75% nitrobenZen! 25% amline 50y0 nitrobenZen! f 50% aniline
Heated 2.5 hr. a t 313' C. Heated 2.5 hr. a t 280' C. Regular Drocedure(368'C. max.) Regular procedure (362OC. niax.)
None
203
206
+
None None None
Nature of Produo Light, fluffy carbon 0.1 g. dense black solid in boat 2.1 g . dark oil in bomb 1.7 g. dense black resin in boat 2 . 7 g,, dense black resin i n boat
Runs 202 and 203 were made in an effort t o determine the effect of heating nitrobenzene for protracted periods below its decomposition temperature. I n sharp contrast t o the analogous runs with nitroxylene, only a small amount of resinous material or dark liquid was found in the boat when the bomb was opened. The liquid left in the boat after run 203 was found by steam distillation t o be at least 7570 nitrobenzene, containing some gummy materials in solution. The remainder of the nitrobenzene in both runs had apparently distilled out of the boat. Since there was no evidence of such a distillation in run 201 where decomposition took place, it seems reasonable t o assume t h a t t h e distillation occurred while the bomb was cooling down and its walls were below the temperature of the sample in the boat. The material recovered from the bomb consisted of water and a n oil. The oil was investigated briefly and was shown to be largely aniline. It seems likely that the metal walls of the bomb catalyzed the hydrogenation of the nitrobenzene after it had distilled out of the boat. Nitrobenzene, therefore, is much more stable thermally than nitroxylene, and most of it survives heat treatment which would condense or decompose nitroxylene. Probably both the methyl groups and the nitro group are involved in the formation of the tarry products produced by prolonged heating of nitroxylene. Runs 205 and 206 show results again differing from those with nitroxylene in t h a t 25% aniline suppressed the explosive decomposition completely. I n these runs also, most of the nitrobenzene appears t o hove distilled out of the boat while the bomb was cooling. Significant quantities of a resinous solid were left in the boat in both runs, however. The amounts were not so large as those obtained with nitroxylene, but were sufficient t o demonstrate t h a t condensation reactions do take place between nitrobenzene and aniline a t elevated temperatures. SUMMARY
1. A method has been developed €or measuring the explosive decomposition temperatures of aromatic nitro compounds. These temperatures depend in p a r t upon sample size and the equipment used. The minimum decomposition temperature of nitrobenzene is about 356" C. and t h a t of nitroxylene is about
308" C. X DlNlTROXYLENE IN MONONITROXYLENE
Figure 3.
Effect of Dinitroxylene on Decomposition Temperature of Mononitroxylene
2. The decomposition temperatures are substantially independent of the pressure of the gas surroundin the samples and are the same in either hydrogen or methane. T i e y are unaffected by the presence of various steels or of active carbon.
1704
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
3. The decomposition temperature of nitroxylene is lowered slightly by the presence of 5 t o 15% of dinitro compounds. 4. Prolonged heating of nitroxylene a t temperatures below its decomposition temperature causes it to resinify. Nitrobenzene is much more stable under such treatment. 5. The presence of 25% xylidine lowers the decomposition temperature of nitroxylene, and a similar amount of aniline prevents the explosive decomposition of nitrobenzene. Mixtures of either of the nitro compounds containing 25% or more of the amine resinify on heating. 6. I n the presence of hydrogen and of molybdenum sulfide on active carbon, the decomposition temperature of nit,roxylene is markedly lowered. This is the result of initial exothermic hydrogenation of the nitro compound.
Vol. 41, No. 8
LITERATURE CITED
(1) Berthelot, -M., Compt. Tend., 105, 1159 (1887). (2) Borsche, W., Ann., 386,359 (1912). (3) Brown, C. L., Smith, FV. hf., and Scharmann, W. G., INU.K N G . CHEM.,40, 1538 (1948). (4) Condit, P. C., Ibid., 41, 1704 (1949). (5) Datta, R. Id., and Chatterjee, N. R., J . Chem. Soc., 115, 1006 (1919). (6) Kolthoff, I. M., and Menzel, H., “Volurrietric Analysis,” t r . b), N. H. Furman, Vol. 11, pp. SO?-11, New York, John wile^. & Sons, 1929. RKB:W~;D April 3, 1948.
Low Pressure Hy rogenation of Nitroxvlene and NiGo over Molvbdenum Sulfide J J
P. C. COSDIT California Research Corporation, Richmond, Calif.
Nitrobenzene and nitroxylene, over molybdenum sulfide on active carbon, may be reduced to the corresponding amines continuously at atmospheric pressure. The catalyst gradually loses activity by becoming coated with tarry materials formed by condensation of the amine with the nitro compound. The hydrogenation of nitrohenzene is improved by raising the pressure to 400-500 pounds per square inch gage. The optimum temperature is above 300” C. and a large excess of hydrogen is desirable. Under these conditions space rates up to 1.5 volumes per volume per hour can be used, and the catalyst is stable. Under the same conditions nitroxylene undergoes some thermal decomposition, and the catalyst eventually becomes fouled with carbonaceous deposits. Batch runs indicate that the catalyst is active down to at least 232” C. at 3000 pounds per square inch gage. This pressure range should be more desirable for the commercial hydrogenation of nitroxylene.
initially formed. In two more recent articles (4)Brown et u i . give data on the use of cobalt,, nickel, cadmium, and lead sulfides as catalysts. Results with these materials were, in general, better t’hanwith t,he previous cat,alysts. The catalytic activity of molybdenum and tungsten sulfidrs has been recognized for many years, and they have been used iru commercial hydrogenations for the manufacture of motor fuels ( 7 , 10, 1%). When the present work was initiated, there was no ret’crenct. to their use for the reduction of nitro compounds. Two recent patents (6, 1 1 ) and a paper ( 2 ) disclose their use for such a purpose. Broyn, Smith, and Scharmann ( 2 )describe a commercial process for the hydrogenation of nitroxylene over molybdenum sulfide on active carbon. They give no laboratory or developmental data, however, and pressures of about 3000 pounds per square inch gage were used. The work to be described here is concerneri almost entirely with pressures of 500 pounds per square inch gag15 or less.
T”’
The materials were the same grades as those described in the preceding article ( 5 ) . Sitroxylene was the purified sample. Tht, rapid method of Tutwiler ( I S ) was used when small concentrations of hydrogen sulfide in gas streams had to be determined. The sulfur cont,ent of liquid products was determined in a Par? bomb. For the routine determination of unreacted nitro coinpoundi in the amines produced by hydrogenation, the following method was used: The sample of organic material was thoroughly dried by shaking with potassium hydroxide pellets, decant,ed through a filter, and shaken again with solid potassium hydroxide. A 10ml. sample of the dried product was pipetted into a 100-ml. conical centrifuge tube. The tube mas filled to the 50-ml. mark wit,h 12% hydrochloric acid, stoppered, and shaken violently. It was then centrifuged unt,il the layers had completely separated. The volume of the lower, organic layer was read, and its color and other properties were observed. Analyses were made in duplicate, and the results expressed as volume per cent of the dried sample. Determinat,ions on synthetic mixtures of aniline and nitrobenzene indicated that the method was good to * 1% of the original sample in the range over which it was used. When the acid-insoluble products were lighter inst,ead of heavier than the aqueous layer, the drtermination was run in a
preceding article (5)discussed the thermal decomposition of aromatic nitro compounds under conditions approaching those of commercial hydrogenation. This paper reports a laboratory investigation of the hydrogenation of these compounds over a catalyst consisting of molybdenum sulfide deposited on active carbon granules. This catalyst has outstanding advantages for commercial use in that it is cheap, light, mechanically rugged, and immune t o poisoning by sulfur, carbon monoxide, and othri impurities often found in commercial hydrogen. The catalytic reduction of nitro compounds has been reviewed in several texts (I, 8). For preparation In the laboratory, Raney nickel or platinum oxide gives excellent results. A large number of other catalysts have been recommended in patents tor commercial use (8); most of these are metals of the copper and iron groups. For the most part, no detailed data are given in the patents on their use. Brown (5) and associates made an extensive study of catalysts for the reduction of nitrobenzene. Most of their work was concerned with metals and metal oxides. I n general, these materials do not appear to be particularly suitable for commercial use since they operate successfully over a limited temperature range and can accommodate only lon- nitrobenzene space rates. Many of them also cause extensive hydrogenation of the amine
!MATERIALSAND A S A L Y T I C A L YJETIIODS
