any point and most within 5%. The value of R2 = 0.954 based on the first- and second-order terms substantiates this relationship. Based on these values the stationary point was calculated as :
Xi =
-0.6621, X2 = -0.1275, and Xj = 0.1818
X3
= -0.3603, X4 = 0.9958,
which yield the input levels as: 4.6 hours, 178' C., 567 p.s.i.g., 2.75 grams of KOH per 5 grams of cellulose and 27/73 water-dioxane ratio. This maximum lies well within the system and has an estimated yield of 1.16 a t that point. A transformation by rotation so that the coordinate axes coincide with those of the quadratic surface yields the canonical form with coefficients : = -0.144044, B22 = -0.08402, B33 = -0.04246, -0.00277, and BE5 = 0.01200
B11
B44
=
From these values it can be seen that the general surface is concave downward about the center with time and temperature as the principal factors. For purposes of further evaluation only these two variables were considered, with the remaining three being held constant a t the stationary point. The equation
Y
= 1.16
- 0.14044 XI'
- 0.08402
~ 2 '
is an ellipse with contour elongated along the x2 axis (temperature axis). The surface for these two variables is relatively
flat in the temperature gradient. Thus, relatively little change in degree of substitution from the optimum occurred between 170' and 180°, but changes in time of one-half hour gave yields which were less than 90% of the maximum. The effect of changes in other variables is negligible in this general region, as they are close to a stationary ridge. The general fit provides an easily adopted scheme for finding the best operating conditions to maximize yield. Conclusions
The vinylation of cellulose in an aqueous system is similar to starch vinylation. The more ordered cellulose structure probably accounts for the somewhat longer time, higher temperature, and higher pressure required for optimum degree of substitution. Time and temperature are the most influential variables. literature Cited
(1) Berry, J. W., Tucker, H., Deutschman, A. J., Jr., IND.ENG. CHEM.PROCESS DESIGN DEVELOP. 2,318 (1963). ( 2 ) Cochran, W. G., Cox, G. M., "Experimental Designs," Chap. 8.4,pp. 335-75, Wiley, New York, 1957. ( 3 ) Siggia, S., Edsberg, R., Anal. Chem. 20, 762 (1948). RECEIVED for review May 5 , 1965 ACCEPTEDDecember 20, 1965 Work done under contract with the U. S. Department of Agriculture and authorized by the Research and Marketing Act. Contract supervised by the Northern Utilization Research and Development Division, Agricultural Research Service.
V A P O R P H A S E N I T R A T I O N OF B U T A N E IN A
MOLTEN S A L T R E A C T O R D E N N I S L . F E A R '
A N D G E O R G E B U R N E T , J R .
Department of Chemical Engineering, Iowa State Uniuersity of Science and Technology, Ames, Iou'a
A molten salt reactor was used to investigate the vapor phase nitration of n-butane. The effects of molten salt temperature, mole ratio of hydrocarbon to nitric acid, and residence time were studied. Nitroparaffin yields, conversions, and product distribution varied widely with molten salt temperatures of 371 ' to 482" C. Product analysis data indicate that the nitroparaffins are produced primarily via the decomposition of alkyl nitrites.
HE VAPOR PHASE XITRATION of propane is a commercially Tsuccessful operation and has been studied extensively (7). Work with butane ( 7 , 5, 7 7 ) is limited, however, in spite of the fact that the same nitroparaffins are obtained from butane as from propane, plus the four nitrobutanes which could be very useful as industrial solvents and chemical intermediates. The cost of butane as a raw material is comparable to that of propane and a t times has been less. One of the reasons that butane has not been as fully investigated as propane is the difficult problem of analyzing the nitration products. Only approximate analyses have been previously reported (2, 4 ) . I n this investigation, highly sensitive gas chromatographic techniques were developed which
Present address, Dow Chemical Co., Midland, Mich. 166
l & E C PROCESS D E S I G N A N D DEVELOPMENT
permit complete quantitative analysis of all significant reaction products (8). As a result, product distribution, yield, and conversion data are available for a range of operating variables, and the mechanism and kinetics of the nitration reactions can be examined. Molten salt reactors have been developed in an effort to provide better temperature control and to avoid the variation of results with time observed in metallic tubular reactors (3). T h e use of nitrate salts prevents gradual poisoning effects from occurring in stainless steel nitration vessels (73, 74),and metal surfaces in contact with molten nitrate salts are very resistant to corrosion by nitric acid ( I ) . For these reasons, a molten salt reactor employing a mixture of sodium and potassium nitrate was used. I n this reactor, the reacting mixture was discharged below the surface of and in direct contact with the molten salt.
Figure 1. A 8
C D E F
G H J
K L
M N P
Q
Schematic flow diagram of the equipment
Glass funnel Graduated acid reservoir Centrifugal pump Filter Air column Rotameter Heat exchanger Butane tank Tank regulator line regulator Rotameter H g manometer Reactor Thermocouple Hg manometer
Description of Apparatus
T h e apparatus consisted of five basic sections: the butane system, the nitric acid system, the reactor, the product recovery system, and the refrigeration system. T h e equipment was modified from that used by Adams ( 7 ) . A schematic flow diagram for the process is shown in Figure 1. Nitric acid and butane were metered through rotameters F and L , respectively. T h e acid concentration was 70 wt.70, and the butane was technical grade with a minimum purity of 95 moleyo n-butane. Prior to entering the reactor, the acid was cooled in a double pipe heat exchanger, G, to help prevent vaporization of the acid before it entered the reaction tube inside the reactor. Vaporization is avoided as boiling nitric acid is corrosive even to stainless steel, and there is evidence that nitric oxide, which is formed during nitric acid decomposition, inhibits nitration reactions (70). Liquid nitric acid also provides some control of the reaction temperature due to heat of vaporization. T h e reactor consisted of a mild steel headplate and cast iron pot. All other parts were of stainless steel. T h e pot contained a molten salt eutectic mixture of 54 wt.% potassium nitrate and 46 wt.% sodium nitrate which melts a t 222' C. The space within the pot was 6 inches in diameter by 12 inches deep. Heat was supplied by three steel-sheathed immersion heaters wrapped around the pot and controlled by variable autotransformers. A cross section of the reactor headplate is shown in Figure 2. Nitric acid entered from the top and sprayed into the reaction tube from four 0.005-inch diameter holes a t the end of the feed line. Here it vaporized and mixed with butane thus initiating
R
s T U
V W X Y
z
-c+
* 8: 0 Q
Product condenser Seporatory funnel l i q u i d product Pyrex condenser Flask Hg manometer Plastic balloon Gloss bottle W a t e r overflow Blunt needle valve Fine needle valve 2 - W a y stopcock 3 - W o y stopcock H g thermometer Bourdon g a g e
the reaction. Air flowing through the annular spaces around the acid tube provided a n insulating barrier to keep the acid cool. A thermocouple measured the acid temperature a t the injection point. Butane entered the reaction tube from a side connection just above the acid spray. Five other thermocouples measured the temperature profile along the reaction tube. From the bottom of the tube the reacting vapors passed through four '/,,-inch diameter holes into the molten salt. Two thermocouples located a t different depths measured the molten salt temperature. T h e product gases flowed from the reactor through condenser R. T h e condensed product was collected in glass separatory funnels, S. Butane and nitrogen dioxide were liquefied in condenser U and collected in a Dewar flask, V . T h e uncondensed gases were collected in a plastic balloon, X,submerged in water. The water displaced by the gases overflowed into a graduated cylinder, Z. Procedure
After temperatures and flow rates had remained constant for about 30 minutes, collection of the products was begun. Temperatures, pressures, and flow rates were recorded every 10 minutes. At the end of a run, liquid product was drained from the separatory funnels in two phases: the first, an aqueous phase containing mostly water and nitric acid; and the second, a n organic phase containing mostly nitroparaffins and oxygenated organics. Samples were analyzed using the gas chromatography procedure developed specifically for this research ( 8 ) . This procedure employs three different columns and two different methods of detection. VOL.
5
NO. 2
APRIL
1966
167
1 t \ 1
NITRIC ACID
'O
A T M O L E R A T I O = 3.4 0 HYDROCARBON YIELD p
z
40
-
E
A M O L E X N E OF T O T A L N P ' S
W
-
PRODUCED
10
-
-
0
114'' INSIDE DlA.
1/16" D l A
Figure 2.
/+
4-HOLES
'
II NO SCALE
H
Reactor headplate cross section 0
Residence time in the reactor was related to the total flow of materials in gram moles per second by locating a thermal conductivity cell a t the reactor outlet. The residence time for a given flow rate was measured by injecting helium into the inlet butane stream and recording the response from the thermal conductivity cell to obtain a n elution curve. The period between the helium injection and the peak of the elution curve (0.5 to 0.7 second for the flow rates used) was taken as the average total residence time. Assuming ideal gas behavior, residence times were calculated to be between 0.02 to 0.04 second in the reaction tube and between 0.04 to 0.07 second in the small vapor space above the molten salt. The sum of these two values, or about 0.1 second, was the time the vapors were not in the molten salt. Conversion and Yield
In this work, the term conversion was defined as the molar ratio of product produced to reactant charged. The term yield was defined as the molar ratio of product produced to reactant consumed. Conversion and yield values for both nitric acid and hydrocarbon were determined for a series of 32 experiments. The data were analyzed using statistical methods to determine the effects of residence time, mole ratio of butane to nitric acid, and molten salt temperature. Nitric acid yields and conversions and hydrocarbon yields were not appreciably affected by changes in mole ratio from 3.4 to 7.8 or by changes in residence time from 0.5 to 0.7 second. Interaction of these variables with temperature also was negligible. The ranges of mole ratio and residence time used were about a maximum for the experimental equipment. T h e data obtained are in agreement with the results of Coldiron et al. ( 6 ) who found that varying the residence time from 0.67 to 1.62 seconds and the mole ratio from 2 to 20 had little effect on conversion in molten salt reactors using propane and nitric acid. Hydrocarbon conversions were significantly decreased by increases in mole ratio because the amount of hydrocarbon present was greatly in excess of that theoretically required. Variations in temperature from 371 ' to 482' C. affected the 168
I&EC PROCESS D E S I G N A N D DEVELOPMENT
c 0
A
Nitric acid conversion. The molar ratio o f nitrogen in the nitroparaffins to nitric acid charged Nitric acid yield. The molar ratio of nitrogen in the nitroparaffins to nitric acid consumed. The quantity of acid consumed is based on the assumption that N O and NOz can b e recovered and converted back to nitric acid Hydrocarbon conversion. The molar ratio o f carbon in the nitroparaffins to total carbon charged for male ratio, butane to acid, of 3.4 only ( 1 2 runs) Hydrocarbon yield. The molar ratio of carbon in the nitroparaffins to carbon consumed, The amount of carbon consumed i s based on the assumption that n-butane, isobutane, and propane are recoverable hydrocarbons Mole% nitroethane of total nitroparoffins produced
results as shown in Figure 3. Since the ranges of mole ratio and residence time were not found significant, conversion and yield values in Figure 3 were averaged a t each temperature for all runs (a total of32). The relative mole% of nitroethane, the most costly nitroparaffin, is also plotted against temperature to show the importance and the difficulty of choosing the most economical operating conditions. As the temperature increases, the mole% of nitroethane increases while yields based on both hydrocarbon and nitric acid decrease. Occurring simultaneously with the decrease in yield was an increase in the quantity of oxidation and decomposition products such as aldehydes, methane, ethane, nitrogen, and oxides of carbon and nitrogen. Conversions of hydrocarbon and nitric acid increase with increases in temperature and then decrease. At 371' C., reaction was incomplete, and some of the nitric acid emerged from the reactor unchanged. At 482' C., nearly 25% of the nitric acid was converted to nitrogen and about 60% to nitric oxide. Product Distribution
Of the CI to ( 2 4 nitroparaffins, only nitromethane, nitroethane, 1-nitropropane, 2-nitrobutane, and 1-nitrobutane were found in significant quantities. Formaldehyde and acetaldehyde were the major aldehydes found along Xvith much smaller amounts of acrolein, propionaldehyde, and n-butyraldehyde.
Table 1. a
I-NP
c.
0 I-NB 0 2-NE
3 a
0
\
AI \
/
Complete Analysis of Reactor Products for Three Selected Runs in Gram Moles Reactor ____ Temperature: -__ O 42 7 482 37 1
I
NM NE 2-XP 2-M-2NP 1-NP 2-NB 1-NB n-C4H1o CH3CHO CzHjCHO CHpCHCHO C3H;CHO
0.04514 0.10510 0.00019 0.00038 0.03366 0.16471 0.08614 4.46626 0.04091 0.00348 0.00197 0.00072 0.00340 Trace Trace 0.30395 0.03202 0,10307 0.66044 0.05419 0.15651 0.14519 0,12844 0.03719 0.01675 0.01987 3.47971 3az
0.03899 0.07377 0.00012 0.00071 0.01075 0.18690 0.04338 5.09020 0.01538
0.00729 TEMPERATURE,
*C
Figure 4. Nitroparaffin distribution as a function of temperature
No alcohols were found except for occasional traces of 1-propanol and 2-butanol. The only ketone found was 2-butanone. The off-gas contained all of the C1 to C4 saturated hydrocarbons, nitrogen, nitric oxide, carbon monoxide, and carbon dioxide. \Vhen nitrations were performed a t 482' C . , traces of hydrogen were found. None of the chromatograms of the off-gas samples showed the presence of olefins although, in a previous investigation where butane was nitrated in a tubular reactor ( d ) , significant quantities of olefins were reported. Perhaps use of the molten salt reactor or the short contact times used in this work prevented the formation of olefins. Complete analyses of nitrator products for selected individual runs a t each of the three temperature levels used are given in Table I. T h e corresponding run conditions are in Table 11. An over-all analysis of the results of all runs shows that. in general, the quantity of oxidation and decomposition productssuch as aldehydes, methane, ethane, carbon and nitrogen oxides-increase with temperature. The ranges of mole ratio and residence time used produced no significant changes in the distribution of nitroparaffin product9. Nitroparaffin distribution data, in moleyo, were averaged for all runs a t each temperature to obtain Figure 4. 2-Nitropropane, 2-methyl-2-nitropropane, and 2-methyl-14tropropane are not included in this figure since they usually were less than 17, of the total nitroparaffins formed. T h e amount of lower nitroparaffins formed increased with temperature. The increase in the amounts of nitromethane and l-nitropropane was small but the amount of nitroethane increased from an average of 20 mole?, a t 371' C. to a n average of 35 mole7, a t 482' C. The amount of 1-nitrobutane increased from an average of 1 2 mole?, a t 371' C. to a n average of 32 mole% a t 482' C. while the average amount of 2-nitrobutane correspondingly decreased from 56 mole% to 9 mole%. This seems to contradict one of the general conclusions drawn by Hass and Shechter (72) that the rate of substitution of the primary and secondary positions approaches equality as the temperature increases. However, the product distribution data probably do not actually represent the rate of substitution between the primary and secondary positions because other factors such as the decomposition rates of 1-nitrobutane and 2-nitrobutane also affect the results. Indicated Reaction Mechanism
T h e product analysis data indicate that the alkyl nitrite
0.51271 0.03552 0.27083 0.50633 0,08162 0.16664
0.02819 3.89156 Run No. a
4c
0.02644 0.07333 0.00018 0.00306 0.02523 0.01283 0.05981 4.29412 0.05482 0.00597 0,00080 0.00119 0.00276 0.22492 0,00728 0.20158 0,94828 0.17497 0.19253 0,07812 0.15028 0.12659 0.02059 4.01129 8b
Obtained by titration oJ aqueous phase.
Table 11.
Experimental Conditions and Results for Three Selected Runs Shown in Table I ___-_Run .Vumber
