Application of Gas− Liquid Film Theory to Base Hydrolysis of HMX

explosive at an initial concentration of 10 wt % PBX 9404 was destroyed in ... characterized for HMX powder and PBX 9404 molding powder from 110 to 15...
0 downloads 0 Views 178KB Size
Ind. Eng. Chem. Res. 1998, 37, 4551-4559

4551

Application of Gas-Liquid Film Theory to Base Hydrolysis of HMX Powder and HMX-Based Plastic-Bonded Explosives Using Sodium Carbonate Robert L. Bishop, Raymond L. Flesner,*,† Philip C. Dell’Orco, Terry Spontarelli, and Sheldon A. Larson Los Alamos National Laboratory, P.O. Box 1663, MS C920, Los Alamos, New Mexico 87545

David A. Bell Department of Petroleum and Chemical Engineering, University of Wyoming, Laramie, Wyoming 82071-3295

Sodium carbonate (Na2CO3) is identified as a hydrolysis reagent for decomposing HMX and HMX-based explosives to water-soluble, nonenergetic products. The reaction kinetics of Na2CO3 hydrolysis are examined, and a reaction rate model is developed. Greater than 99% of the explosive at an initial concentration of 10 wt % PBX 9404 was destroyed in less than 5 min at 150 °C. The primary products from Na2CO3 hydrolysis were nitrite (NO2), formate (HCOO-), nitrate (NO3-), and acetate (CH3COO-) ions, hexamethylenetetramine, (hexamine: C6H12N4), nitrogen gas (N2), nitrous oxide (N2O), and ammonia (NH3). The rate of hydrolysis was characterized for HMX powder and PBX 9404 molding powder from 110 to 150 °C. The rate was found to be dependent on both the chemical kinetics and the mass transfer resistance. Since the HMX particles are nonporous and external mass transfer dominates, gas-liquid film theory for fast chemical kinetics was used to model the reaction rate. Introduction The Department of Defense and the Department of Energy dispose of large quantities of high explosive (HE) material, including unexploded ordnance and bulk explosives recovered from aging munitions stockpiles.1 Traditionally, open burning/open detonation(OB/OD) is used to destroy these materials. However, concern over dispersion of harmful constituents produced during OB/ OD operations has prompted the development of safe and environmentally benign alternatives. One HE disposal process being developed by Los Alamos National Laboratory,2-10 industry,11-13 and elsewhere14-19 is base hydrolysis. Base hydrolysis breaks down nitramine explosives into organic and inorganic salts, soluble organic compounds, benign nitrogen gases (primarily nitrous oxide (N2O), and nitrogen (N2)), and ammonia (NH3). The process operates at relatively low pressures and temperatures compared to those of OB/OD, is easy to implement and control, and is conducted in a closed system. An ammonia-based hydrolysis process is operational at Thiokol’s Bringham City, Utah, plant to treat double-based rocket propellant.13 In addition, Chemical Systems Division, a propellant manufacturer in northern California, is using base hydrolysis to destroy waste energetic materials. A pilot-scale unit is also soon scheduled for implementation at the Department of Energy’s Pantex Plant. The objective of this research is to investigate the effectiveness of Na2CO3 as an agent for hydrolysis of HMX (1,3,5,7-tetraaza-1,3,5,7-tetranitrocyclooctane) (see Figure 1) and PBX 9404 (94% HMX, 3% nitrocellulose, 3% chloroethyl phosphate, and 0.1% diphenylamine). †

Fax: (505) 667-0500. E-mail: [email protected].

Figure 1. HMX molecule.

Currently, the most commonly used base in explosive hydrolysis is sodium hydroxide (NaOH). Base hydrolysis using NaOH at or above 90 °C is an effective method for the disposal of many explosives. Lower material cost is the chief advantage of using Na2CO3 instead of NaOH. At present, NaOH is $0.33/kg20 for a 50 wt % aqueous solution. Sodium carbonate is $0.12/kg.21 This difference results in a saving of 28% on a molar basis. Treatment with 1.0, 1.5, or 2.5 M Na2CO3 at temperatures above the normal boiling point of water (110-150 °C) converted the explosives into nonexplosive, aqueous compounds. The fraction of solids converted to soluble compounds and the experimental time-temperature profile were used to determine reaction rates. Aqueous reaction products were NO2-, HCOO-, NO3-, CH3COO-, NH3(aq), and hexamine. The gaseous products were N2, N2O, and NH3. Both chemical kinetic and mass transfer resistances were found to be significant, and a model combining both was therefore required. Gas-liquid film theory for a fast irreversible reaction was used to derive the model. The model predicted the conversion of HMX or PBX 9404 to within an average of 10%. Experimental Methods Base hydrolysis was carried out using HMX crystals and PBX 9404 molding powder. The HMX crystals

10.1021/ie980351a CCC: $15.00 © 1998 American Chemical Society Published on Web 11/06/1998

4552 Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998

Figure 2. 100 mL reactor schematic.

consisted solely of coarse particles with a mean diameter of 234 µm. PBX 9404 is composed of 94 wt % HMX, 3 wt % nitrocellulose (NC), and 3 wt % tris(β-chloroethyl) phosphate (CEF). The HMX contained in the PBX 9404 consisted of coarse particles with the same 234 µm mean diameter (calculated multipoint BET surface area of 0.13 m2/g)22 and fine particles with a mean diameter of 5 µm (calculated multipoint BET surface area of 1.8 m2/ g).22 These particles were blended in a coarse-to-fine ratio of 3:1 by weight. Molding powder consists of HMX particles that are surrounded by the binder (NC) and plasticizer (CEF). The mixture agglomerates into granules with a size range of 1-3 mm. Molding powder can then be pressed and machined to desired geometries for use in munitions. Two reactors were used for the kinetic studies. Most of the experiments were carried out in a 100 mL Hastelloy “C” (nickel alloy C276) reactor (Figure 2) that was heated and stirred with a hot plate/stirrer (VWR Scientific series 400HPS). The temperature and pressure readings were monitored and read into a computer through an IEEE-488 signal by a Stanford SR630 thermocouple monitor (Stanford Research Systems). A computer using a data acquisition program (Labview, National Instruments) recorded this signal at 5 s intervals. The reactor was heated to a set point temperature and held for several minutes and then rapidly quenched in an ice bath. One sample port was used to take gas samples and to vent the off-gas at the end of each experiment. The entire apparatus was placed in a fume hood behind an explosive blast shield. The second reactor (Figure 3) was a 2 L stainless steel reactor (Parr Instrument Co.). Electric coil heaters in contact with the reactor walls heated the reactor and an internal cooling coil using tap water provided cooling. A PID controller (Watlow Series 945) maintained a set temperature within 5 °C. The data acquisition program (LabView, National Instruments) recorded temperatures and pressures every 5 s. A stirrer shaft with two flat impellers continuously mixed the explosive powders in the Na2CO3 solution. The rotational speed of the stirrer was also recorded every 5 s. A gas sampling/ vent port and a liquid sampling port enabled liquid and gas samples to be removed during experiments. The temperature control system of the 2 L reactor quickly regulated the temperature rise due to heat generated during hydrolysis by internal cooling, whereas the 100 mL reactor did not have this feature

Figure 3. 2 L reactor schematic.

and the set-point temperature was exceeded by as much as 25 °C during experiments. Both reactors were batch loaded and sealed at the beginning of each experiment. After reaction, the liquid was filtered using a Buchner funnel with an 8 µm cellulose filter paper. The solid residue was dried on the filter paper under vacuum for at least 1 h and then weighed to determine mass conversion. Gas and liquid samples were taken at the completion of reactions. Gas samples were taken by venting the reactor head gas into a 100 mL stainless steel sample chamber. Liquid samples were poured into glass sample vials after the solution was filtered. Additional samples were periodically collected during reactions in the 2 L reactor. Compositions of the gas samples were determined using gas chromatography (Hewlett-Packard Series II 5890A gas chromatograph). The gas chromatograph used both a thermal conductivity detector (TCD) and a flame ionization detector (FID) in parallel. Three columns were used: Supelco 2-5461 Carboxen 1006 Plot, Supelco 2-5463 molecular sieve 5A plot fused silica, and Chom pack plot fused silica with Poraplot Q coating. Liquid samples were analyzed for pH (Orion 290A pH meter), ion content (Dionex ion chromatograph analyzer) and total organic and inorganic carbon (Rosemount Dohrmann DC-190 total organic carbon analyzer). The ion chromatograph used a CS14 column with electrochemical suppression and conductivity detection. The carbon product distribution was also analyzed by NMR (JEOL JNM-GSX270 FT-NMR). Results and Discussion Base Hydrolysis Experiments. Experiments were performed varying the initial Na2CO3 concentration, reaction temperature, time the solution was held at the reaction temperature, and explosive loading. The range of values of these parameters is shown in Table 1. In a few experiments, the agitation speed was also varied. Over 35 experiments were performed. With a PBX 9404 solid loading of 0.1 g/g liquid, a 1.5 M Na2CO3 solution at 150 °C destroyed 99% of the explosive in 5 min. The pressure in the reactor due to the accumulation of gaseous products increased as high as 21 bar in the 100 mL reactor and 5 bar in the 2 L reactor. Pressure had no noticeable effect on the reaction rate.

Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998 4553 Table 1. Range of Important Experimental Parameters for the HMX Powder-PBX 9404 Kinetic Study parameter

HMX expts

PBX 9404 expts

temp (°C) base concentration (M) solid loading (g of explosive/ g of liquid) time at temp (min)

120-154 1.0-1.5 0.05-0.3

110-143 1.0-1.5 0.07-0.3

3-18

4-30

[HMX]ppm ) 1.04 × 100.025T(°C)

The solution pH was 11 ( 1, below the Resource Conservation and Recovery Act (RCRA) corrosive level (pH < 12.5). Table 2 summarizes the aqueous and gaseous reaction products for a typical experiment at an explosive loading of 10 wt % in solution and 1.5 M Na2CO3. Of the carbon remaining after reaction, 46% was HCOO-, 9% was CH3COO-, and 45% was not detected by either ion or gas chromatography. The 45% of unknown compounds likely consisted of CO32-, hexamine, and formaldehyde breakdown products as suggested by NMR (Figure 4). The 35% unknown of the total nitrogen was most likely made up of NH3, aqueous hexamine, and other amines as indicated by previous studies.2,3,14-16,23 A carbon-13 NMR scan of the products from 1.5 M NaOH hydrolysis of HMX is also presented in Figure 4. In contrast to NaOH-mediated hydrolysis, few aqueous products were produced: mainly CO32-, HCOO-, NO3-, NO2-, NH3(aq), and hexamine. Reaction Rate Modeling. Determination of OHConcentration. Previous research has determined that approximately 4 mol of hydroxide ion (OH-) is consumed per mole of HMX,23 (one hydroxide for each nitro group of the compound) (see Figure 5). Hydroxide ions form in two reactions of CO32- with water and in the dissociation of water (eqs 1-4). The OH- concenK1

CO32- + H2O 798 OH- + HCO3K1 )

[OH-][H2CO3-] (1) [CO3]

K2

HCO3- + H2O 798 OH- + H2CO3 K2 ) K3

[OH-][H2CO3] (2) [HCO3-]

H2CO3 798 CO2(g) + H2O K3 ) K4

PCO2 [H2CO3]

H2O 798 OH- + H+ K4 ) [OH-][H+]

HMX Solubility. Temperature and ionic strength of the base solution influence HMX solubility. The effect of temperature (T) determined empirically by Kramer and Kuehne26 is given in eq 5. A relationship between

(3)

HMX solubility and ionic strength is unavailable. Thus the change in solubility due to increased ionic strength of the solution was not incorporated in the model. However, a trend toward lower solubility at higher ionic strength has been reported.3 A model was developed to predict the influence on reaction rate of Na2CO3 concentration, temperature, HMX explosive loading, and mixing conditions (agitation, baffling, reactor size, impeller size). Before modeling, the appropriate reaction regime had to be determined. If kinetically limited, the intrinsic chemical kinetics can be used. If mass transfer limited, fluid mechanical conditions must be included. To determine the proper regime, experiments varying the mixing conditions were performed. Reaction Regime. At 140 °C, when the Reynolds mixing number was increased from 75 to 600 by increasing agitation, the reaction rate increased 7.5%. In an identical experiment using a NaOH solution, in which the rate has been shown to be mass-transfer limited above 70 °C,6,7,9 the rate increased 75%. The smaller increase in rate for Na2CO3 hydrolysis suggested that the liquid mass transfer rate did not significantly influence the Na2CO3 hydrolysis. To determine surface area effects, the overall rates for the hydrolysis of HMX-coarse (mean particle diameter dp ) 234 µm) and HMX-fine (mean particle diameter dp ) 5 µm) were compared. However, conclusive results were not obtained because the smaller particles agglomerated in solution. Since the reaction rate appeared to be independent of stirring rate, a model incorporating only chemical kinetics was developed.27 Previous studies show that the hydrolysis of HMX can be represented by a secondorder rate equation, first order in aqueous HMX concentration and first order in OH-.28-30 The second-order rate constant is then described by the Arrhenius relation. The results from Heilmann et al.16 are shown in eq 6.

-RHMX

(

)

mol‚L ) min

(

5.9659 × 1017 exp (4)

tration was calculated as a function of temperature and ionic strength from thermodynamic data24 and an activity coefficient model developed by Pitzer.25 Since the equilibrium constant for eq 1 is at least 2 orders of magnitude greater than those for the other equations generating OH-, this reaction determines the OHconcentration. Therefore, although each Na2CO3 molecule is capable of generating 2 OH- ions, the actual value is somewhat lower due to the small equilibrium constants, K1, K2 and K4. Since eq 1 controls the OHconcentration, it was assumed that 4 mol of CO32- were converted to HCO3- and OH- per mole of HMX. The values of the equilibrium constants at standard temperature, K298, are listed in Table 3.

(5)

)

-13459K [HMX]aq[OH-] (6) T (K)

The activation energy constants determined from the Na2CO3 experiments were inconsistent with previous results28-31and were approximately 60% of the value determined by Heilmann et al.16 Therefore, the assumption of a kinetically limited reaction inadequately described this system, and a new model incorporating both mass transfer and kinetic rates was developed. Application of Gas-Liquid Reaction Rate Model. The model was derived from gas-liquid boundary layer theory for a fast reaction.32-38 Extension of the gasliquid boundary layer theory to solid-liquid systems presents no problems since in each case reaction occurs completely within the liquid phase and the gas or solid simply supplies reactant to the liquid. The model assumes that the reaction is contained within a bound-

4554 Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998 Table 2. Product Analysis for Na2CO3 Base Hydrolysis of HMX Powder and PBX 9404 Molding Powder with Results from an Average of Five Experiments Using Different Experimental Conditions concn (ppm)

% tot. C or N in explosive processed

Carbon in Solution 2700 ( 300 14,000 ( 1300

inorg carbon org carbon formate (aq) acetate (aq) carbon monoxide/carbon dioxide carbon (not detected by IC or GC)

17 ( 2 86 ( 8

Carbon-Bearing Species 33000 ( 11000 2800 ( 1500 trace

44 ( 15 9(5 trace 47 ( 16

Nitrogen-Bearing Species 22000 ( 2000 trace 1400 ( 100 26000 ( 8000

nitrite (aq) nitrate (aq) nitrogen (gas) nitrous oxide (gas) ammonia (gas + aq)

19 ( 2 trace 3.8 ( 0.3 46 ( 14 ∼30a

Other hydrogen a

Predicted from previous

trace

trace

work.3,6,8,14,16,23

Figure 5. Reaction of HMX with sodium carbonate. Table 3. Equilibrium Constants and Heats of Formation for the Principle Reactions Governing the Concentration of Hydroxide Ions in Solution for Na2CO3 Hydrolysis with Constants Evaluated at 298.15 K reacn

K298

∆H (kJ/mol)

1 2 3 2+3 4

2.1 × 10-4 2.4 × 10-8 30 7.1 × 10-7 1.0 × 10-14

41 48.3 20.3 68.6 55.8

Figure 6. Boundary layer schematic for fast reaction. COH and CHMX are the aqueous concentration of OH- and HMX, respectively. CHMX goes to zero in the boundary layer while COH is unchanged. δ is the boundary layer thickness. Figure 4. 13C nuclear magnetic resonance results for the base hydrolysis of HMX powder using sodium carbonate and sodium hydroxide. Experiments were performed using the 100 mL Hastelloy “C” reactor with 50 mL of solution and a 2.54 cm stir bar rotating at 400 rpm. Experimental conditions: 150 °C, 1.5 M base solutions, and a solid loading of 0.1 g of HMX/mL of base solution.

ary layer surrounding the HMX particles (Figure 6). The resulting expression for sufficiently fast, irreversible, power law kinetics with dissolution is shown in eq 7,33,35,36 where DHMX is the HMX diffusivity in water, a

is the interfacial surface area per unit reactor volume, kH is the chemical kinetics rate constant determined by Heilmann et al.,16 (see eq 6), and n is the reaction order for HMX.

x

-RA,obs ) a

2DHMXkH[OH-] [HMX](n+1)/2 (n + 1)

(7)

This equation shows, for a sufficiently fast reaction, the rate varies with the square root of kH rather than

Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998 4555

ΘV a)R dp

(9)

shrinking ball model with constant particle population was used to describe the decrease in particle size during hydrolysis. An empirical constant, R, is only a function of the particle geometry but also accounts for the interactions between particles, such as agglomeration, that reduce the wetted surface area. For a sphere that is completely wetted, R is 6. By substituting for a using eq 9, the final form of the rate expression is shown in eq 10.

-RHMX )

Figure 7. Plot of Hatta number as a function of reaction temperature. The diffusivity was determined by the Wilke-Chang equation. The rate constant was determined by Heilmann. The liquid mass transfer rate, kL, was calculated using a correlation developed for agitated vessels by Kawase and Moo-Young.

kH. Therefore, the apparent activation energy is half of the chemical kinetics activation energy. This qualitatively agrees with the measured activation energy for heterogeneous HMX hydrolysis (66.5 kJ/mol), being approximately 60% of the chemical kinetics activation energy (111.9 kJ/mol).16 The rate equation shows a direct dependence of rate on the interfacial surface area of HMX particles, a. The predicted rate is also independent of the mixing conditions, which was observed experimentally. A sufficiently fast reaction can be defined as that having a Hatta number33,34,36,37 greater than 4. The Hatta number (eq 8) is a dimensionless expression that

Ha ∝

chemical reaction rate ≡ mass transfer rate

x

δ

2kH[OH-]

(n + 1)DHMX



x

2DHMXkH[OH-] (n + 1)kL2

(8)

compares the chemical kinetic rate to the mass transfer rate. In eq 8, δ is the boundary layer thickness. For Hatta numbers greater than 4, the chemical reaction is much faster than the rate at which reactants travel through the solution, and the reaction is contained within a boundary layer surrounding the HMX particles. To calculate Hatta numbers for the HMX hydrolysis reaction, the Wilke-Chang39 equation was used to determine DHMX and the mass transfer coefficient, kL, was determined using the procedure developed by Kawase and Moo-Young40 for an agitated vessel. The kinetic rate was determined using a method developed by Heilmann.16 Figure 7 shows that the Hatta number is greater than 4 for reaction temperatures above ∼100 °C. Below ∼100 °C, the reaction rate is so slow that it is controlled by chemical kinetics. Since the Na2CO3 hydrolysis reaction must be carried out at temperatures above 100 °C to achieve practicable reaction rates, the Hatta number is greater than 4 and eq 7 was used. The interfacial surface area, a, in eq 7 can be written in terms of mean particle diameter, dp, and the volume fraction of solid explosive in the slurry, ΘV (eq 9).41 The mean particle diameter varies with reaction time. A

R Θ D k [OH-]HMXaq dp Vx HMX Heilmann

(10)

Values of R were determined using eq 10 and measured rates under several different experimental conditions. The modeling equations were numerically integrated over time using the Eulerian method. Reducing the sum of square residuals optimized the value for R. The calculated values for R were 2 for HMX in the 100 mL reactor (correlation coefficient r ) 0.91, number of points n ) 22, 95% confidence interval on r (0.79 < F0.025 < 0.965) and 4 for HMX in the 2 L reactor (r ) 0.998, n ) 5, 0.56 < F0.025 < 1). The discrepancy between R for HMX in the 100 mL and 2 L reactor is due to the difference in mixing behavior in the two systems. HMX crystals tended to clump together and some of the smaller crystals remained on the surface of the liquid. In the 100 mL reactor, stirring was performed by a magnetic stirrer bar sitting at the bottom of the reactor. This did not break up the clumps or help wet the HMX crystals as well as the turbine stirrer used in the 2 L reactor. For most applications of base hydrolysis of HMX, the HMX will be in the form of a plastic-bonded explosive rather than pure crystals. The usefulness of a rate model is therefore increased if it is shown to predict destruction rates for HMX-based explosives. Experiments were carried out to destroy PBX 9404 molding powder. Experiments showed that the molding powder began to disintegrate in an agitated basic solution at 70 °C, and the individual granules were completely broken apart within minutes. Since the hydrolysis rate of HMX is extremely slow at 70 °C, it appeared that, by the time a temperature was reached at which the hydrolysis rate became significant, the molding powder was broken up into individual HMX particles. If it is also assumed that the NC and CEF either completely hydrolyze or that NC and CEF do not influence the HMX reaction, then eq 10 can be applied to PBX 9404 molding powder. Experiments showed that in a pressed piece of PBX 9404, the HMX granules are not released into solution as readily. Therefore, a representative value of dp could not be determined, and the model is limited to molding powder only. The HMX particles in PBX 9404 molding powder, however, are not identical in size to the HMX crystals. Skidmore et al.42 have shown that the formulation process reduces the size of the HMX particles by approximately 20%. The mean diameter, dp, of a 3:1 fine to coarse particle blend of HMX is reduced during formulation from 175 to 134 µm. The R value for HMX crystals in molding powder was 13 (r ) 0.91, n ) 17, 0.78 < F0.025 < 0.965). The value of R for PBX 9404 is

4556 Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998

Figure 8. Parity plot comparing the developed film theory model to experimental results. These data cover the complete experimental parameter range as given in Table 1.

13 for both the 2 L and the 100 mL reactor. For the PBX 9404, the molding powder agglomerates are large enough (1-3 mm) to prevent “floating,” and wetting did not appear to be a problem. The difference in value for the HMX and PBX 9404 molding powder is likely due to differences in surface morphology between the HMX coarse and fine crystals, differences in agglomeration and wetting behavior, and error associated with determining a mean particle diameter for PBX 9404 and HMX. A parity plot comparing the model to experimental results for HMX crystals and PBX 9404 molding powder is shown in Figure 8. Good agreement is shown, with the predicted conversion within an average of 10% of the result for experiments using 1.0 and 1.5 M Na2CO3. Figure 9 shows how the reaction conversion varies with temperature. Duplicate experiments were performed with four different temperature set-points; the time temperature curves are shown in Figure 10. The only significant deviation of the model from the experimental data is at 150 °C. At 150 °C, the reaction kinetics may have been fast enough so that the mass transfer of OHinto the boundary layer was significant. This would lower the overall experimental reaction rate, causing the model to over predict the explosive conversion. To further investigate the performance of the model, transient samples were taken during three experiments, and the concentrations of the two principal product ions, formate and nitrite, were determined. Liquid samples were taken during each experiment and then rapidly cooled in ice to stop further reaction. A comparison of the predicted and measured ion concentrations for HMX hydrolysis is shown in Figure 11. The effect of temperature on the PBX 9404 hydrolysis rate is shown in Figure 12. Increasing the temperature from 130 to 150 °C enhanced the rate by a factor of 5. At 150 °C, over 99% of the PBX 9404 degraded after 3 min, whereas, only 30% reacted after 20 min at 100 °C. The Na2CO3 concentration and mass loading might also influence the model accuracy. Figure 13 shows the effect of both mass loading and base concentration. The

Figure 9. Comparison between experimental data and model estimates of reaction conversion. Four different temperature set points were used. Base concentration was 1.5 M and 10 wt % mass loading was used. Each experiment was duplicated. The explosive was HMX with a mean diameter of 234 µm.

Figure 10. Experimental temperature-time profiles used for Figure 9.

model matches experimental data well for 1.0 and 1.5 M Na2CO3 but significantly overestimates the conversion at 2.5 M. The initial rate of PBX 9404 base hydrolysis at 2.5 M Na2CO3 was 29% of the rate at either 1.0 or 1.5 M. Spontarelli et al.23 reported a similar observation during NaOH hydrolysis; the overall rate decreased with NaOH concentration above 1.5 M. This was believed to be caused by the decrease in HMX solubility due to an increase in the ionic strength of the solution. Since dissolved HMX carries no charge, an increase in the solution salt concentration or ionic strength will decrease HMX solubility. The current model predicts the conversion of HMX or PBX 9404 to within an average of 10 wt % for 1.0 to 1.5 M Na2CO3 solutions. For the model to fit the 2.5 M Na2CO3 data,

Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998 4557

Figure 11. Comparison between predicted and experimental nitrite and formate ion concentrations as a function of time. Three different experiments are plotted. Experiments differed in both mass loading (10-7 wt %) and set-point temperature (150-130 °C). The explosive was HMX with a mean diameter of 234 µm. Figure 13. Effect of major parameters on the model accuracy. Average values of actual and predicted conversions for several experiments are plotted. The explosive was PBX 9404.

at 0.05 g/g is most likely from a reduction in particleparticle interactions, which would increase the interfacial surface area. Figure 14 shows predictions for the concentration of both the reactants and products based on an experimental temperature profile. The consumption values of CO32- and HMX solid are plotted along with the production of N2O, N2, HCOO-, NO2-, and CH2COO-. The plot also shows the interfacial concentration of HMX, OH- concentration, temperature, pressure, and mean particle diameter. The plot shows that the OHconcentration increases with temperature. Since eq 1 is endothermic, an increase in temperature increases the equilibrium constants K1. The formation of the product species, N2, N2O, formate, and nitrite, were modeled using a constant product to HMX conversion ratio as determined from the results in Table 2. The plot shows that the concentration of HMX solid decreases rapidly after the temperature reaches 120 °C. The CO32- decreases at 4 times the rate of HMX due to its consumption through the reaction. The concentration of aqueous HMX is highly dependent on temperature and increases rapidly above 100 °C. Finally, the plot shows the mean diameter reduction with time. Figure 12. Effect of temperature on conversion rate, predicted using a theoretical, isothermal temperature profile. Base concentration was 1.5 M and explosive loading was 10 wt %. The explosive was PBX 9404 molding powder.

the empirical constant, R, must be reduced to 4 to account for lower HMX solubility. The model correlates well with experimental data over the mass loading investigated in this study, with an exception at low loading. The deviation from the experimental results

Conclusions The kinetics of the base hydrolysis of HMX and PBX 9404 using Na2CO3 were investigated. At a temperature of 150 °C, a base concentration of 1.5 M, and a 10 wt % explosive loading in the solution, 99% percent of all the solids were destroyed in less than 5 min. Fewer reaction products formed when Na2CO3 was used as the hydrolyzing agent compared to NaOH. The final solu-

4558 Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998

Figure 14. Modeling prediction for HMX (mean diameter of 234 µm) hydrolysis with 1.5 M Na2CO3 with a temperature set point of 150 °C for 3 min and overall reaction time of 30 min. The overall conversion of HMX was 72% from the model and 69% experimentally. The consumption of carbonate was 4 mol/mol of HMX. The molar ratio of products to HMX was 1.4 for nitrite and formate, 1.7 for nitrous oxide gas, and 0.18 for nitrogen gas.

tion pH for Na2CO3 hydrolysate was below the (RCRA) corrosive level (pH < 12.5). Na2CO3 is a feasible alternative to NaOH. Na2CO3 should also be effective in other plastic-bonded explosives in which the majority of the explosive compound is HMX. The hydrolysis of HMX and PBX 9404 using Na2CO3 was modeled using boundary-layer theory for a fast, irreversible reaction. When using 1.0-1.5 M Na2CO3, the model predicted HMX conversion to within an average of 10% of experimental results. Both PBX 9404 molding powder and HMX powder hydrolysis follow the same rate law. Only an interfacial surface area constant, R, is adjusted for the differences in the average HMX particle diameters. Acknowledgment We acknowledge the DOE/DoD Memorandum of Understanding Agreement for funding this research. We also thank Joe Repa and Stephen Newfield for their support and guidance. Special thanks go to Rhonda McInroy, John Kramer, Ed Eaton, Michael Hiskey, and Jose Archuleta for their assistance and helpful discussions. Literature Cited (1) Moore, T. Rising to the Challenge in Military-Site Cleanups. Environ. Eng. World 1995, 1, 28. (2) Skidmore, C.; Dell′Orco, P.; Flesner, R.; Kramer, J.; Spontarelli, T. Chemical Destruction of HMX-Based Explosives with Ammonium Hydroxide. LA-13003l; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1995. (3) Spontarelli, T.; Buntain, G.; Sanchez, J.; Benziger, T. Destruction of Waste Energetic Material Using Base Hydrolysis. 12th Incineration Conference, Knoxville, TN, 1993; University of California: Irvine, CA, 1993; pp 787-793. (4) Buntain, G. A.; Sanchez, J. A.; Spontarelli, T.; Benziger, T. M. Destruction of Waste Energetic Material Using Base Hydroly-

sis. LA-UR 93-1527; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1993. (5) Dell’Orco, P.; Flesner, R.; Spontarelli, T.; A., S. J.; Kramer, J. F. Base Hydrolysis and Hydrothermal Processing of Composition B-3 Explosive. LA-12960-PR; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1995; Vol. 1. (6) Flesner, R. L.; Spontarelli, T.; Dell’Orco, P. C.; Sanchez, J. A. Base Hydrolysis and Hydrothermal Processing of PBX 9404. ACS Symposium Series, Emerging Technologies in Hazardous Waste Management VI; Atlanta, GA, 1994; American Chemical Society: Washington, DC, 1994. (7) Flesner, R. L.; Spontarelli, T.; Dell′Orco, P. C.; Sanchez, J. A. Base Hydrolysis and Hydrothermal Processing of PBX 9404 Explosive. LA-UR-95-795; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1995. (8) Flesner, R. L.; Dell’Orco, P.; Bishop, R. L.; Spontarelli, T.; Uher, K. Pilot-Scale Base Hydrolysis Processing of HMX-Based Plastic Bonded Explosives. LA-UR-96-1798; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1996. (9) Flesner, R.; Bishop, R.; Dell’Orco, P. Pilot Scale Base Hydrolysis of PBX 9404. Global Demil Symposium, Sparks, NV, 1996; Joint Ordnance Commanders Group/American Defense Preparedness Association of JOCG/ADPA: 1996; pp 606-620. (10) Sanchez, J. A.; Kramer, J. A.; Dell′Orco, P. C.; Spontarelli, T.; Flesner, R. L. Base Hydrolysis and Hydrothermal Processing of PBX 9404 explosive. LA-UR-95-1074; CONF-941255-3; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1994. (11) Borcherding, R. An Alternative to Open Burning Treatment of Solid Propellant Manufacturing Wastes. Waste Manage. 1997. 17, 135-141. (12) Borcherding, R. Treatment of Solid Propellant Manufacturing Wastes, Base Hydrolysis as an Alternative to Open Burning. JANNAF Safety and Environmental Protection and Propellant Development and Characterization, San Diego, CA, 1995. (13) Cannizzo, L.; Mower, G.; Hunstman, L.; Achatz, W.; Edwards, W. Hydrolysis of the Energetic Materials Present in a Composite, Modified, Double Bass Propellant. Conference on Alternatives to Incineration, 1995. (14) Heilmann, H. Physico-Chemical Treatment of Water Contaminated with the High Explosive RDX and HMX Using Activated Carbon Adsorption and Alkaline Hydrolysis. Ph.D. in Civil Engineering and Environmental Engineering Thesis, University of California at Los Angeles, 1996. (15) Heilmann, H. M.; Weismann, U.; Stenstrom, M. K. Kinetics of the Alkaline Hydrolysis of High Explosive RDX and HMX in Aqueous Solution and Absorbed to Activated Carbon. Environ. Sci. Technol. 1996. 30, 1485-1492. (16) Heilmann, H. K.; Stenstrom, M. K.; Hesselmann, R. P. X.; Wiesmann, U. Kinetics of the Aqueous Alkaline Homogeneous Hydrolysis of High Explosive 1,3,5,7-Tetraaza-1,3,5,7-Tetranitrocyclooctane (HMX). Wat. Sci. Technol. 1994, 30, 53-61. (17) Bunte, G.; Krause, H. Disposal of Energetic Materials by Alkaline Hydrolysis and Combined Techniques. Int. Annu. Conf. ICT (Energetic Materials: Insensitivity and Environmental Awareness), 1993. (18) Bunte, G.; Hirth, T.; Krause, H. Pressure Hydrolysis of Energetic Materials-New Possibilities for Explosive Disposal. Int. Annu. Conf. ICT 1993, 24th, 35-1/35-15. (19) Bunte, G.; Krause, H.; Hirth, T. Disposal of Energetic Materials by Alkaline Hydrolysis and Combined Techniques. Propellants, Explos., Pyrotech. 1997, 22, 160-167. (20) Dow, sodium hydroxide price quote from Dow sales, Sept 1995. (21) FMC, sodium carbonate price quote from FMC sales, Sept 1995. (22) Archuleta, J. Particle Size, BET Surface Area, and Pore Distribution for HMX-F 82C000E094 and HMX-C H83-L030-50. Lab Report; Data Sheet 4/11/97; Los Alamos National Laboratory: Los Alamos, NM, 1997. (23) Spontarelli, T.; Buntain, G. A.; Flesner, R. F.; Sanchez, J. A.; Unkefer, P. J. An Engineering System Using Base Hydrolysis for Complete Destruction of Energetic Materials. International Symposium on Energetic Materials Technology, Clarion Plaza Hotel, Orlando, FL, 1994; American Defense Preparedness Association (ADPA): 1994; pp 55-61. (24) Lide, D. R. Handbook of Chemistry and Physics, 67 ed.; CRC Press: Cleveland, OH, 1987; pp D51-D93.

Ind. Eng. Chem. Res., Vol. 37, No. 12, 1998 4559 (25) Pitzer, K. S. Activity Coefficients in Electrolyte Solution, 2nd ed.; CRC Press: Boca Raton, FL, 1991. (26) Kramer, J. F.; Kuehne, T. Base Hydrolysis Modeling. presentation; Amarillo, TX, 1995. (27) Bishop, R. L.; Flesner, R. L.; Dell′Orco, P.; Spontarelli, T.; Bell, D.; Kramer, J. Base Hydrolysis Kinetics of HMX-based Explosives using Sodium Carbonate. LA-UR-96-1818; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1996. (28) Epstein, S.; Winkler, C. A. Studies on RDX and Related Compounds VI. The Homogeneous Hydrolysis of Cyclotrimethylenetrinitamine (RDX) and cyclotetramethylenetetranitramine (HMX) in Aqueous Acetone, and Its Application to Analysis of HMX in RDX(B). Can. J. Chem. 1951, 29, 731-733. (29) Jones, W. H. Mechanisms of the Homogeneous Alkaline Decomposition of Cyclotrimethylenetrinitramine: Kinetics of Consecutive Second- and First-order Reactions: A Polarographic Analysis for Cyclotrimethylenetrinitramine. J. Am. Chem. Soc. 1954, 76, 829-835. (30) Hoffsommer, J.; Kubose, D.; Glover, D. Kinetic Isotope Effects and Intermediate Formation for the Aqueous Alkaline Homogeneous Hydrolysis of 1,3,5-Triaza-1,3,5-trinitrocyclohexane (RDX). J. Phys. Chem. 1977, 81, 380-385. (31) Croce, M.; Okamoto, Y. Cationic Micellar Catalysis of the Aqueous Alkaline Hydrolysis of 1,3,5-Triaza-1,3,5-trinitrocyclohexane and 1,3,5,7-Tetraaza-1,3,5,7-teranitrocyclooctane. J. Org. Chem. 1979, 44, 2100. (32) Bird, B. R.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomenom; 1st ed.; John Wiley and Sons: New York, 1960; pp 532, 620. (33) Bell, D.; Bishop, R. Application of Gas Liquid Film Theory to Fast Liquid Phase Reactions of Sparingly Soluble Solids. To be published.

(34) Vazquez, G.; Chenlo, F.; Peeereira, G. Enhancement of the Absorption of CO2 in Alkaline Buffers by Organic Solutes: Relations with Degree of Dissociation and Molecular OH Density. Ind. Eng. Chem. Res. 1997, 36, 2353-2358. (35) Van de Vusse, J. G. Mass Trasnfer with Chemical Reaction. Chem. Eng. Sci. 1960, 16, 21-30. (36) Hikita, H.; Asai, S. Gas Absorption with (m, n)-th Order Irreversible Chemical Reaction. Int. Chem. Eng. 1964, 4, 332340. (37) Prisciandaro, M.; Pepe, F. Adsorption with Zero and Pseudo-zero Order Chemical Reactions. Can. J. Chem. Eng. 1997, 75, 362368. (38) Astarita, G.; Marrucci, G. Gas Abdorption with Zero-order Chemical Reaction. Ind. Eng. Chem. Fundam. 1963, 2, 4-7. (39) Reid, R. C.; Praunitz, J. M.; Sherwood, T. K. The Properties of Gases and Liquids, 3rd ed.; McGraw-Hill Book Co.: New York, 1977; pp 590-595. (40) Kawase, Y.; Moo-Young, M. Solid-Turbulent Fluid Heat and Mass Transfer: A Unified Model Based on the Energy Dissipation Rate Concept. Chem. Eng. J. 1987, 36, 31-40. (41) Treybal, R. E. Mass-Transfer Operations, 3rd ed.; McGrawHill: New York, 1980; pp 522-533. (42) Skidmore, C. S.; Phillips, D. S.; Son, S. F.; Asay, B. W. Characterization of HMX Particles in PBX 9501. LA-UR-97-2596; Los Alamos Unclassified Report; Los Alamos National Laboratory: Los Alamos, NM, 1997.

Received for review June 4, 1998 Revised manuscript received September 5, 1998 Accepted September 6, 1998 IE980351A