I
JULIUS ROTH, F. S. STOW, Jr., and D. L. KOUBA Research Center, Hercules Powder
Co., Wilmington, Del.
Kinetics of Ethylene Glycol Nitration This reaction is of great importance in the explosives industry, which uses ethylene glycol dinitrate as a sensitizing ingredient for dynamites
THE
over-all thermochemistry of the nitration of polyhydric alcohols by mixed nitric-sulfuric acids is fairly well understood, but reaction rate data are lacking. As part of a program to develop a continuous method of nitration, the kinetics of nitration of ethylene glycol was studied under steady-state flow in a tubular reactor. As aromatic nitrations have been extensively studied (4,comparison of the kinetics of aromatic and alcohol nitrations should be interesting.
allowed to separate into “oil” and acid phases. The spent acid phase (SA) was analyzed for sulfuric acid, nitric acid, water, and dissolved ethylene glycol dinitrate. Samples were also taken to check the flow rate. The main effluent stream was drowned in ice water. Recovery of ethylene glycol dinitrate from the drowning water showed that with mixed acids of less than 15% water, conversion of glycol to ethylene glycol dinitrate was quantitative. FLOWMETER
Procedure
Proportioning pumps were used to deliver constant amounts of ethylene glycol (G) and mixed acid (MA). The two streams impinged on each other at the tee (Figure 1) and flowed down a stainless steel tube l / g inch in inside diameter under turbulent conditions. A portion of the tube next to the tee was insulated ; the remaining length was chilled to prevent excessive solubility of ethylene glycol dinitrate (EGDN) in spent acid. Along the outside of the tube, but under the insulation, thermocouples were positioned to follow the course of the reaction. Because the two streams mix rapidly and the reaction itself is rapid, conditions in the reactor portion were essentially adiabatio. The system was calibrated by pumping sulfuric acid against water. The observed heats in these experiments checked closely with accepted literature values for the heat of dilution of sulfuric acid (7). During actual nitration runs samples of effluent were taken and
Results
The mixed acid ranged in composition from 11.5 to 20% water, 18 to 40% nitric acid, and 45 to 68.5% sulfuric acid. The ratio of mixed acid to ethylene glycol weight ranged from 12 to 24 (Table I). Good material balances were obtained for almost all the experiments. Only runs having good material balance were used for kinetic analysis. Typical temperature-time records FLOWMETER ACID ( O O C . )
GLYCOL (20OC.)
50 CM.
+=
u
50 CM.
= THERMOCOUPLE STATION
c 50 CM.
REACTION TUBE 0.309 CM. 1.D. (0.117 IN. 1 x 470 CM. LONG 50 CM. 50
* CM.
50 CM.
L
t TO DROWN Figure 1. Ethylene glycol and mixed acid impinged on each other at the tee VOL. 50, NO. 9
SEPTEMBER 1958
1283
.-.-.-. R U N NO.10
II.5 O/o H20
3\1 32
’
12A
0.;
I
0.;
’
’
’
0.6 I 0.8 TIME, SEC.
’
’
I.o
1.2 I
Figure 2. Heat evolution is strongly influenced by amount of water in the mixed acid
’
O M O L E RATIO
H2S04 TOTAL ACID
RATIO
H2S04 T O T A L ACID
.MOLE
(Figure 2), for comparabIe suIfuric acidnitric acid and total acid-ethylene glycol ratios, show that heat evolution of the reaction is strongly influenced by the amount of water in the mixed acid. The bottom curve of Figure 2, over the time range studied, has practically zero slope. I n this experiment no ethylene glycol dinitrate was recovered on drowning. I t is believed that the initial AT in this case is due to the heat of solution of glycol in mixed acid, followed by a slow nitration. This belief was confirmed by several experiments, in which glycol was added to “watery” mixed acids as well as to aqueous sulfuric acid (Figure 3). I n all these calorimeter experiments appreci-
Table 1.
H2S04
I 0.8
1
0.4
able temperature rises were observed and the final solutions were homogeneous. I t was also qualitatively shown that sulfonation is slow compared to nitration. In the experiment corresponding to the second curve, about 50% conversion of glycol to ethylene glycol dinitrate was
Initial Concentrations Immediately after Mixing Tee
0.82
i
1.00
1
1
1.2
1.6
HzSOa
“01
HzO
Ethylene Glycol
9 13
10.93 9.09 10.91 9.74 10.16 10.50 8.60 6.52 7.60 10.27 10.44 9.85 10.41 9.43 10.03 10.38 10.55
4.97 4.71 4.96 4.81 4.92 4.41 7.30 9.57 4.31 4.92 4.92 5.35 4.82 6.02 4.61 4.78 4.85
10.00 16.50 9.97 14.31 12.77 13.22 12.80 12.56 18.81 12.90 12.06 12.77 10.79 11.78 9.97 10.38 9.98
1.74 1.81 1.76 1.68 1.70 1.50 1.47 1.25 2.84 1.50 1.58 1.62 2.03 1.87 1.73 1.30 1.06
Mean Feed Temp., O
c.
5.2 2.9 3.5 1.6 3.0 4.5 4.4 12.5 2.5 3.7 3.7 5.3 -0.1 4.7 4.4 3.1 3.7
2.0
INDUSTRIAL AND ENGINEERING CHEMISTRY
2.4
H2° TOTAL ACID
Heat Balance for Complete Nitration
Before making a kinetic analysis of the reaction. the observed maximum heat rise for reactions in which nitration was complete must agree with the heat rise calculated from the thermochemical data. The net reaction is
+ MA = ( mEGDN + HNOala + {SA + (1 - m ) EGDN)b AHT
Phase alpha consists of ethylene glycol dinitrate and any nitric acid which is dissolved in ethylene glycol dinitrate. Phase beta consists of spent acid and ethylene glycol dinitrate dissolved in spent acid. This net reaction may be broken down into M.4 2Hz0 (liq.) - 2HN03 (liq.) =
+
SA1 AHd
+
(nHN0s)o G f 2HN03 (liq.) = EGDN SA1 = ( S A ) @
AH,
+ 2I-120
(liq.) AHrL
EGDN = (d3GDN)a
+ EGDN)p ((1 - m) AH,
and AHTr = AHd Jr A&
1 284
I
\ I
observed. For the other two curves, conversions were quantitative.
G
Run
35 39A 39B
ii
Heat of solution of glycol in mixed acids
(Moles/liter)
15 16 17 18 19 20 21 22 23 24 25 26
0.64
MOLERATIO
Figure 3.
\’\
i ;
4- AH, 4-
ETHYLENE GLYCOL N I T R A T I O N Table
II.
Observed and Calculated
AT,,, Agreed for Experiments Where Nitration Was Complete
RUn
No.
9 15 19 20 22 23 24 25 26 35 39A 39B
"t c
g 0.8
! I , , , , , ,
8 O.,
Obsd. 37.1 39.5 26.8 22.0 27.8 34.2 31.5 43.8 35.8 36.8 27.6 23.4
38.1 39.2 25.8 20.9 28.7 31.3 30.4 41.8 35.3 38.1 27.5 22.3
,
s 2
I
0
0.5 E N T ACID COMPOSITION, UOLE RATIO
Figure 4.
a. b
I
0.8
O.?
0.9
OF NITRIC ACID/TOTAL ACID
1.0
groups, and AHH,is the heat of solution of glycol in an acid corresponding to a degree of nitration o f f . I n the early stages of reaction AHQ (obtained by interpolation from Figure 3) provides most of the observed heat but becomes relatively unimportant a t f > 0.5.
Heat of solution of ethylene glycol dinitrate in spent acid
(all values at 18' c.). A H , is the heat of dilution of the acids, including the heat necessary to abstract the nitric acid for the nitration from the mixed acid. I t is calculated from the'integral heats of solution of mixed acids and nitric acid given by Rhodes and Nelson ( 5 ) . AH,, the heat of nitration, is obtained from heats of formation given by Rossini and others (7). Measurements were also made to obtain rough values of AH, and AH8. These determinations are believed to be of sufficient accuracy, because the contributions of these two heat terms are small. Figure 4 shows AH8 for different acid compositions. AH, was determined to be about zero. All AHT's were corrected to the maximum temperature observed in the corresponding experiment , In every case AHT was calculated to be lower than that observed if AH,, is 3.5 kcal. per mole [based on the literature value (7) of the heat of formation of EGDN]. (AHobs - AHd - AH8)/ moles glycol = constant = 8.5 f 0.3 kcal. per mole. AHn in (7) is based on heat of combustion of ethylene glycol dinitrate obtained by Rinkenbach (6). Small amounts of unnitrated organic impurity can raise the heat of combustion of ethylene glycol dinitrate sufficiently to make the reported value of the heat of formation of ethylene glycol dinitrate too low (Rinkenbach's ethylene glycol dinitrate contained 18.37% nitrogen compared to the theoretical 18.42%). Based on a rough redetermination of the heat of combustion of pure ethylene glycol dinitrate in a Parr bomb, AH,, was found to be about G kcal. per mole of glycol. It was decided to use AH, = 8.5 kcal. per mole of glycol, as determined above, as this value is believed
to be more reliable than that obtained from the Parr bomb. Calculated and observed values for AT,,, are compared in Table 11. Agreement is generally good.
Calculation of f as a Function of Time
With the above assumptions, f can be obtained as a function of time from the observed temperature-time records. AT', the temperature rise at any f, is AHT', corrected to actual reaction temperature, divided by the heat capacity of a mixture of extent of reaction f. Plots of AT' us. f are linear, with a few minor deviations. The slopes and intercepts of these plots are shown in Table 111. With these AT' us. f relationships and from the measurements of AT' as a function of time (Figure 2 ) f is readily determined as a function of time. Typical f us. t plots are shown in Figure 5.
Heat Balance for Partial Nitration
If the reaction proceeds in stages: glycol (G) + glycol mononitrate (MG) +- ethylene glycol dinitrate (EGDN), for partial nitration it is necessary to know the heat terms due to mononitrate. If one assumes, however, that any heat terms due to mononitrate are the average of the heat terms due to ethylene glycol dinitrate and glycol, it can be shown that AHT' = A H d ' $- fAH,, $- AH8' $(1 - f ) AHo, where the primes indicate partial nitration, f is extent of reaction = esterified OH groups per initial OH
Table 111.
Slopes of ATcalad.vs. f and log 1 /(1
iiTo,Icd ./f t
Run No. 9 13 15
16 17 18 19 20 22 23 24 25 26 35 39A 39B
O
c.
18.8" 17.2 17.5c 15.4 16.4 15.5 15.8 14.8 13.4 16.1 14.31 18.6 17.9 16.4 11.5
10.0
l / t Log
Intercept, O
c.
21.0 13.8 21.4 15.4 16.8 14.0 10.5 6.0 15.4 16.0 16.0 23.2 17.4 21.7 16.0 12.6
- f)
vs. f plots
Sec.-1
Slope of Log Plot is Constant in this Range off
14 0.028 14 0.160 0.76 0.52 1. 54, 2.27e 1.35 2.39 1.63 2.700 1.72 6.45 2.0w 5.6
0.8 to O.gb 0 to 0.2b 0.8 t o Q . g b 0 to 0.3' 0 t o 0.8 0 to 0.6 0.5 to 0.9 0.60 t o 0.95 0.25 t o 0.9 0.1 to 0.9 0 to 0.5 0.4 to 0.9' 0 to 0.9 0.4 t o 0.85b 0.5 to 0.9' 0.4 to 0.85'
[ l / ( l - f)]
Slope slightly less at f < 0.4. No experimental determinations outside of this range. Slope decreases for f < 0.5. Slope slightly greater at f > 0.86 and slightly less at f < 0.3. Slope decreases forf < 0.6. f Wide scatter of hTcalod. 8 Shows small positive intercept.
VOL. 50, NO. 9
SEPTEMBER 1958
1285
and I
I.OC
dr
-ky
& = 2x
t
where k = k 5 / k 4
0.8-
on integrating, 0.6-
x
0.4-
and 0.2-
0.1
0.2
0.3
‘’
0 4
0.5
0.7
0.b
0.8
Substituting these quantities into the equation definingf
1.0
0.9
T I M 6 . SEC.
Figure
5. Extent of reaction
is increased in concentrated mixed acid
I
Numbers on curves are run numbers
If k
f is zero a t a small finite time. Presumably this time lag is the time of mixing of acid and glycol. There is a definite trend to smaller time lags with increasing linear flow rate through the reaction tube. This is to be expected, as more turbulent flow and better mixing will result a t higher flow velocities. As shown in Figure 6, log 1/(1 - f) is directly proportional to t over a wide’ range off’s. Here t has been corrected for the small time lag due to mixing. Slopes of all log 1/(1 - f) us. t plots are listed in Table 111.
(1 -f) =
or
f= OH groups esterified OH groups initially present
+ NOz++
k4
Relationship o f f to Glycol Concentration. Defining the following quantities as:
MG
MG
+ H+
+ N O z ”ks4 EGDN + H +
then
a = (G)o
x = (G),
23
/
. * , 0 I
0.1
Figure 6.
0 7
Log 1 I(1
I
0. 3 0.4 TIME, SEC.
’
0.5’
I
0.6
- f ) is directly proportional to time f = extent of reaction Numbers on curves are run numbers
1 286
-y
+2a 22
Assuming for the moment that nitronium ion is the nitrating agent (the argument holds for whatever nitrating species one assumes, if the same species is doing the nitrating in both cases) : G
C
1,
then
Discussion
I
-
INDUSTRIAL AND ENGINEERING CHEMISTRY
t
As log (1/1 - f) = Ct, where C is the slope for each experiment of the linear plots of Figure 5, the rate of disappearance of glycol i s first order with respect to glycol. If k is different from unity, at least in the early stages of reaction, Ct = A log (a/.) log B where A and B are constants different for different ratios of ks/kd. Mechanism of Nitration of Glycol. CHOICEOF NITRATING AGENT. That log ( a / x ) = 2 Ct, with C constant for a given set of initial conditions but varying several hundredfold between different experiments (Table 111), indicates that either the nitrating species is relatively constant throughout a run, or that the drop in rate due to depletion of nitrating agent is compensated for by an increase in temperature as the reaction proceeds. Chedin (7), in studying the various components present in mixed acids, concluded that the concentration of NOa+, nitronium ion, becomes close to zero for acids containing more than 13% water. His work relied on Raman spectra to follow NOz+ and it seems likely that his method is insensitive for determining small concentrations of NOz’. Chedin’s work does not preclude the possibility of the existence of small amounts of NOz+ in mixed acids which initially contain up to 17% water. I n addition to NO*+, undissociated nitric acid and Hz+N03 are possible nitrating agents. The nitric acid concentration, however, is rapidly reduced as the reaction proceeds, and an unduly high temperature effect would have to be invoked to explain the pseudo-first-order behavior mentioned above. Because the present data cannot distinguish between NOzf and H z + N O ~ , because the kinetic expressions would
+
ETHYLENE GLYCOL N I T R A T I O N differ only by a constant, it is assumed that NO*+ is the nitrating agent. MECHANISM OF NITRATION. The very rapid initial temperature rise observed is due to solution of glycol in the mixed acid. Thereafter the nitration proceeds as a homogeneous reaction with ethylene glycol dinitrate continuously separating from the acid phase. From the following reaction sequences : 1 . Hn0 H2S04-Ha+O HS04Ki 2. H2O "Os-"03. HnO K2 3. HzS04 +HNO,-H2O NOz+ HSO4Kt
+ +
1.0
+
+
+
+ N O S + + MG + H + 5. MG + NO2+ EGDN + HS rapid 6 . H f + HzO H3+0 k4
4. G
k5
+
Figure 7.
it follows that
kq
exhibits expected variation with temperature
or from Equations 1 and 5 because steps 1, 2, 3, and G, being essentially acid-base reactions, are fast compared to 4 and 5 . Assume that water is first used up in completely ionizing sulfuric acid; any water left over then hydrates nitric acid (2, 3) and the water formed during nitration is immediately converted into Ha+O. If
'
x
(G)t
a
(G)a
b = (HN0a)o c =
(Hz0)o
OH groups esterified OH groups initially present
F = fraction of HNOa hydrated (2), then, at any extent of reaction f, from Equation 3
(1
since NOz+ will be very small and F approaches a constant value, F*, when C - d is greater than 1.1. ( b - 2af)
+
M
const. (1
- f), which
is the ob-
served relationship of Equation 2. If (c - d) is small but still positive, F is also small, and neglecting NOz+ in comparison with ( b - 2Qf)
7 50
- f)
-
- f)
const. (1
(8)
If d is greater than c but (d - c) is small Equations 7 and 8 apply. Although the sulfuric acid is initially in excess, here it is assumed that the water formed by equilibrium (3) will ionize all the sulfuric acid, and NO2+ is still small compared to ( b - 2uf). In Equations 6 and 8 the term ( b 2 4 ) (d -k 2af) decreases mildly with increasing f. At the same time k4 is expected to increase with increasing f, because the temperature is rising. These compensating effects could account for the observed constancy of Equation 2.
Now
(d 2uf)(b - 2uf) does not vary by more than a factor of 2 in the range off = 0.2 to f = 0.8 and the error resulting from neglecting 2 4 , at small f, and F* ( b - 2uf) at large f,in the denominator will be relatively small. Therefore,
df dt
d = (H2SOi)o f =
and
r
.Y
x-x0-0
(4)
Equation 4 is not easily solved for NO2+, so the limiting cases will be considered first. Where c is appreciably larger than d, Equation 4 may be approximated by
El INPUT ACID, LB. IMIN. OUTPUT ECDN, LB. /MIN.
I
I NITRATING ACID COMPOSITION
(d
+ 2af)Kb - 2 a f ) ( l - F)1 - d - F(b - 2uf)J* [C
H2S04 HN03 EGDN H20
(5)
as NO2+ is expected to be very small. According to the above reaction sequence --dG _ -= -dx
dt
dt
I 1.0
66.32 22.770
i,a% 9.2%
I
I
I
2.0
3.0
4.0
REACTION TIME, S E G
2k4(.~)(NO2 +)
Figure 8.
Data from larger equipment duplicate those from smaller system VOL. 50, NO. 9
SEPTEMBER 1958
1287
Determination of k4 for Experiments Where (Hs~O),2 (HzS04),
Table IV.
(Allconcentrations in moles/liter) f = 0.8
f = 0.5
(HaO), ( H Z S O ~ ) ~ (HT;Os)b
Temp.,"
c.
Run 136 16' 17 18 19 20 22 23 24 25 26 35 39A
20.0 18.6 28.1 26.5 19.8 27.0 25.8 27.7 28.5 32.3 30.1 34.4 24.8
7.41 4.57 2.61 2.72 4.20 6.04 2.63 1.62 2.92 0.38 2.35 .0.06 0.0
'
0.65 1.08 1.02 0.77 2.25 3.16 1.20 1.89 1.29 2.47 2.08 2.88 3.48
~
Temp.,"
(NOz")'
2kcd
0.009 0.115 0.31 0.21 0.38 0.31 0.34 0.57 0.35 0.72 0.53 0.83 0.85
3.1 1.4 2.5 2.5 4.0 7.3 4.0 4.2 4.4 3.7 3.3 7.8 2.4
O
c.
...
23.01 31.30 30.9 27.6 30.3 29.8 32.0
...
38.0 36.5 39.3 28.3
(HaO), (HzSOc)o
...
4.57 2.68 2.72 4.20 6.04 2.63 1.62 . e .
0.38 2.35 -0.06
0.0
(NOS')'
2kad
0.91 0.86 0.49 1.51 2.54 0.59 1.00
0.085 0.22 0.10 0.26 0.27 0.15 0.41
1.9 3.4 5.2 5.9 8.4 9.0 5.8
1.26 1.07 1.85 2.70
0.56 0.34 0.70 0.78
4.7
...
...
...
...
" Obtained from AT v s . f plots. ("01) = (b - 2u.f) (1 - F ) : not corrected for NO$' formed. (d + 2uf) ("03) - (Noz') = Slope of log 1/(1 - f) v s . * equation solved by successiveapproximations. Ka '(N02') { NOa' -t c - d - F ( b - 2 ~ f ) ) ~ ' (NOe').
" f = 0.1.
'f = 0.2,
g /
4'. G
+
5'. MG
"03
--+
+ HNOa
----t
+ HzO EGDN + HzO
replace sequences 4 and 5 , then fort 2 d, df = k'p (6
dt
- 2af)(1 - $')(I - f)
where S' N const. = slope of ChCdin, FCntant, and Vandoni data ( 2 ) . ( b 2 4 ) is constant only at very smallf. For c>>d, df - k'c ( b - 2 a f ) ( 1 - F*)(l - f) 3 -
Here again ( b - 2uf) is approximately constant only at smallf. Calculation of Constants. If K1/K3 is known and is assumed to be independent of temperature, NOz+, for any f, can be estimated by use of Equation 4 by successive approximations. Dividing the slope of any log 1/(1 - f ) us. t plot by the appropriate NOz+ gives a n estimate of 2 kq for the particular f for which N 0 2 + was computed. Because f can be related to temperature (Table 111), a n estimate can be made of the variation of kd with temperature. In principle it is possible to obtain 2 kd &/Kl from Equation 5 and 2 kq (K8/KJ113from Equation 7 . Ir is then possible to get K1/& from the quotient of these two values if the comparison is made a t about the same temperature. Unfortunately, data for the regions where Equations 5 and 7 are applicable are scanty and estimates of K1/K3 vary from about 30 to about 125 moles per liter. Considering all the data and using Equation 4 as outlined above, it was found that KI/Ka = 50 moles per liter gave the best fit. The calculations are summarized in Table IV. Figure 7 is a least square plot of log 2 kr us. I/T. Although the scatter is large,
1288
(moles)(liter ) -I( sec. )-I
MG
k '6
...
5.1 9.2 2.6
t plots divided by
=0.7.
If sequences k'4
...
where the uncertainties are for the 95% confidence level, it appears that this treatment reduces the observed several hundredfold variation of pseudo firstorder rate constants to within a factor of 2 or 3. The frequency factor and activation energy are of the expected order of magnitude, and KI/& is consistent with the assumption that any water in the mixed acid is first used in ionizing the sulfuric acid. kd is not greatly influenced by the choice of K1/K3. For K I / K ~= 125 moles per liter the k4)s are about 1.5 times larger than at K1/K3 = 50, but the activation energy is very close to that given above. The large scatter is probably due to differences between numbers of the same magnitude which have to be used in the computations. If there is a large excess of nitric acid kd is about twice as large as the average. No explanation can be offered for this observation.
0.5-Inch Reactor
A final step in this program was to determine whether or not the conditions of rapid mixing followed by rapid but controllable nitration couId be successfully carried over to a large scale. To this end a reactor was constructed of stainless steel tubing 0.5 inch in inside diameter. Flow control and mixing were essentially the same as in the smaller (0.117-inch) reactor. A %foot long adiabatic reaction section was followed by an 80-foot-long cooling-coil section immersed in cold brine. I n this way reaction conditions could be followed in the adiabatic section but still the effluent ethylene glycol dinitrateacid mixture was cooled enough to handle and separate according to the normal commercial practice.
INDUSTRIAL AND ENGINEERING CHEMISTRY
Nitration runs have been made over a wide variety of acid compositions and flow rates. Temperature-time data for two typical runs are shown in Figure 8. The data from this large-scale equipment, while less complete, duplicate in every way those from the more fully instrumented smaller system. I t is believed that the problems of still further scale-up are small and that the way has been shown to a new, safe, controllable method for the continuous nitration of glycol, glycerol, and related materials. Acknowledgment
The writers thank W. E. Davis for doing most of the thermochemical computations and for help in the derivation of some of the equations. L. G. Bonner made several helpful suggestions in the interpretation of results. Literature Cited
(1) Chedin, J., Chim. @ ind. (Paris) 61, -571 (1949). (2) Chgdin, 'J., FBnBant, S., Vandoni, R., Compt. rend. 226, 1722 (1948). (3) Gillespie, R. J., Hughes, E. D., Ingold, C. K., J . Chcm. Sod. 1950,. p. _ 25%. ' (4) Hughes, E. D., Ingold, C . K., Reed, R. I., Zbid., 1950, p. 2400.
(5) Rhodes, F. H., Nelson, C. C., IND. ENG.CHEM.30, 650 (1938). (6) Rinkenbach, W. H., Zbid., 18, 1195 (1926).
(7) Rossini, F. D., Wagman, D. D., Evans, W. H., Levine, S., Jaffe, I., "Selected Values of Chemical Thermodynamic Properties," Natl. Bur. Standards, Circ. 500 (1952). RECEIVED for review January 4, 1958 ACCEPTED May 13, 1958 Divisions of Chemistry, Technology, Acid Esters, York, N. Y.,
Carbohydrate and Cellulose Symposium on Chemistry, and Pharmacology of Nitric 132nd Meeting, ACS, New September 1957.
