Elif Irem Senyurt,Mirko Schoenitz,Edward L.Dreizin

New Jersey Institute of Technology,Newark,NJ,07102,USA

Keywords:Chemical weapon agent Agent defeat Thermal decomposition Incineration Prompt reactions

ABSTRACT Rapid destruction of stockpiles of sarin and other chemical weapon agents(CWA)requires understanding and quantitative description of the relevant chemical reactions.Rapid reactions at elevated temperatures are of particular interest for prompt agent defeat scenarios.Diisopropyl methylphosphonate(DIMP)is a sarin surrogate particularly well suited to model sarin thermal decomposition and is often used in experiments.This article is a review of different experimental methods addressing decomposition of gasified DIMP,respective results and their interpretations.Major early decomposition products are propene,methylphosphonic acid,methyl(oxo)phosphoniumolate,and isopropanol.Early computational work using available kinetic data for fluorine and the phosphorus-fluorine bond predicted the decomposition under incineration conditions.Experiments using an isothermal flow reactor operated at 700-800 K were used to model DIMP decomposition as unimolecular reaction with results that were consistent with the earlier theoretical work.Decomposition in dynamic environments was studied using DIMP supported on rapidly heated substrates.The results showed different decomposition products and product sequences forming at different heating rates,suggesting the need for revised reaction kinetics.However,species quantification in such experiments is difficult because of inherent large temperature gradients.Plasma produced in a corona discharge was also reported to lead to rapid DIMP decomposition at low temperatures.Decomposition products were distinct from those observed at high temperatures.Shock tube experiments may be well suited to study decomposition of organophosphorus compounds like DIMP following their rapid heating in diverse environments.However,presently,only sarin surrogates other than DIMP have been investigated,and no intermediate reaction products,important for developing a validated mechanism,could be detected.

1.Introduction

1.1.Chemical weapon agents and prompt agent defeat

Chemical warfare is the use of the toxic chemical substances to kill,injure or incapacitate an enemy[1,2].A chemical substance intended for such use is a chemical weapon agent(CWA).North Atlantic Treaty Organization(NATO)classified CWAs as blister agents,nerve agents,asphyxiants,choking agents and incapacitating/behavior altering agents[3-5].

CWAs pose a threat to humankind and are widely banned[6,7].Therefore,guidelines for environmentally safe destruction of CWA stockpiles need to be developed[7,8].Because CWAs are highly toxic,direct experimental studies are limited to those conducted in specially equipped military laboratories,and research on CWAs is often conducted using simulant compounds.An ideal CWA simulant(CWAS)would imitate the pertinent chemical and physical properties of the agent without its associated toxicological properties[3,4].

Efforts focused on safe and effective destruction of CWA stockpiles have been active for decades[1].Most available CWA decontamination methods require exposure times of minutes or hours.Such methods are useful when decontamination chemicals and equipment can be delivered and operated directly at the location of CWA stockpiles or infected areas[5,9].Conversely,challenges arise when access is limited,and CWA needs to be destroyed rapidly,without unintended release,or other collateral effects.For such cases,technologies ensuring prompt defeat of CWAs[10],with characteristic times of milliseconds to seconds become crucial.

Means of prompt defeat must be developed for destruction of both chemical and biological agent stockpiles,especially in cases when CWA aerosols or gases can be released due to elevated temperatures[11].Indeed,high temperatures typically occur in fireballs,generated to neutralize or destroy the agents.The associated blast overpressure risks spreading the CWAs before they are decomposed[10,11];such effects should be minimized.Therefore,respective munitions should rely on reactive materials generating copious amounts of heat without substantial gas release[12,13].To develop effective prompt defeat techniques and respective reactive materials,detailed measurements and models need to first focus on laboratory experiments with CWAS.The CWAS must be reproducing the CWA behavior in conditions involving rapid heating and exposure to high temperatures for durations typical of fireballs.This poses constraints on both,the choice of CWAS,and the experimental approach for studying their decomposition reactions.

Reported methods for destroying CWAs can be broadly split into three categories:thermal decomposition[14,15],chemical degradation[16]and catalytic decontamination[4,17].Thermal decomposition is achieved either by incineration[18]or pyrolysis[15,19].Chemical degradation reduces the toxicity of CWAs by alkaline solutions and other oxidants[1,16,17,20].Catalytic decontamination turns CWAs into benign chemicals using catalysts[21,22].Chemical degradation,commonly involving reactions in liquid phase,is a relatively slow process[20,23];therefore,it may not be suitable for prompt defeat applications.The other two methods,thermal decomposition and catalytic decontamination can occur in the gas phase and thus are important for understanding and describing various CWA prompt defeat processes depending on the temperature and materials generated by the fireball.Because elevated temperatures always accompany a fireball,thermal decomposition is always expected to be a process important for prompt defeat.

There are several reports in the literature on rapid thermal decomposition of CWAS and CWA.For example,Zegers et al.[15],describe the gas phase pyrolysis of diethyl methylphosphonate(DEMP),which is a surrogate for the nerve agent sarin.More recently,Shan et al.[24],studied the thermal decomposition of sarin.Both studies described a significant reduction of the agent or surrogate amounts after a short,sub-second exposure to elevated temperatures.Data describing intermediate products formed in such reactions are limited,however.Some of such intermediates may remain harmful,and thus it is important to develop a comprehensive mechanism describing their formation and lifetime.In a practical situation associated with expanding and decaying fireballs,reaction times may vary from a fraction to hundreds of ms and the temperature can rapidly change from hundreds to thousands of K.The oxidizing environment may also change,from oxygen starved to fuel-lean.Studies aimed to support prompt defeat methods should thus determine how such dynamically changing environments affect CWA decomposition,what the intermediate products are,and how rapidly they decompose into harmless species.

1.2.Diisopropyl methylphosphonate(DIMP)as CWAS

Nerve agents are considered to be the most nefarious[5,25]among all CWA types,and they became major components of modern arsenals[26].These CWAs are organophosphonates,which are organic substances containing a phosphoryl(P=O)or thiophosphoryl(P=S)groups giving them a phosphorylating action mechanism[5].Sarin is a dangerous CWA that belongs to the nerve agent class.Exposure is lethal even at very low concentrations[5].Diisopropyl methylphosphonate (DIMP), dimethylmethylphosphonate (DMMP), diethyl-methylphosphonate(DEMP),triethylphosphate(TEP)and trimethyl phosphate(TMP)are common simulants for sarin.The structures of sarin and its common simulants are shown in Fig.1.

Of the different simulants shown in Fig.1,DIMP may be the most promising for studying thermal decomposition mechanisms of sarin[27,28].DIMP is the only one with the same oxygen/isopropyl group(-O-C3H7)as found in sarin.The only structural difference to sarin is the substitution of the second oxygen/isopropyl group with fluorine[1,16].All other CWAS molecules in Fig.1 have different ligands for both,the oxygen/isopropyl and fluorine groups.The structural similarity between DIMP and sarin is expected to lead to similar decomposition mechanisms,helping one establishing more accurate sarin decomposition models[28].

A theoretical work by Glaude et al.[29],modeled the incineration of sarin and its surrogates DIMP,DMMP,and TMP at high temperatures(800-2200 K)with the exposure time of 0.1 s.Natural gas fired incineration was simulated with a mixture of natural gas,air and 0.1% of CWA or its surrogates under perfectly stirred reactor conditions.The computational fluid dynamics code Aurora with the Chemkin model and kinetic database[30],including the rate of unimolecular DIMP decomposition proposed based on experiments[14],were used to predict reaction rates and consumption of the reactants.Selected calculated results are illustrated in Fig.2.The same plot shows recent experimental data for sarin[24].In experiments,sarin was vaporized with nitrogen serving as carrier gas.The gas mixture was further diluted with nitrogen,preheated to the reaction temperature and sent to a temperaturecontrolled tube furnace.Sarin pyrolysis was studied at temperatures in the range of 350-500°C(623-773 K)with residence times of 0.062-0.5 s.The product gas mixture was collected using a liquid nitrogen trap and the remaining sarin was analyzed by gas chromatography.To compare experimental results from Ref.[24]to calculations from Ref.[29],the experimental data were extrapolated to the calculated 0.1 s residence time as shown in Fig.2.A reasonable agreement is observed between the predicted and observed behaviors for sarin.It is also observed that predicted behaviors were quite similar to each other for thermal decomposition of DIMP(solid line)and sarin(dashed line)under typical incineration conditions.Other simulants,TMP and DMMP,were predicted to be consumed at much lower rates[29],making them less suitable choices as CWAS for studies of thermal decomposition of sarin.

Fig.1.Chemical structures of sarin and its common simulants.

Fig.2.Computed sarin,DIMP,DMMP and TMP decomposition by incineration at different temperatures with residence time of 0.1 s[29]compared with experimental sarin decomposition[24].

In separate recent experiments[31],rapid pyrolysis of DMMP and DIMP adsorbed to a porous substrate was achieved using laser heating.It was found that some 10% of the DMMP remained unreacted when the nominal substrate temperature reached 2300 K,whereas nearly all DIMP decomposed with the substrate heated to a nominal～1200 K.Thus,consistent with earlier models[29],the decomposition rate of DIMP is greater and is expected to approach that of sarin,unlike for the more slowly decomposing DMMP.

Based on the above results,it is concluded that DIMP is the most suitable simulant for sarin,especially when its rapid thermal decomposition is of interest.Therefore,this report will focus on thermal decomposition of DIMP.Issues warranting future studies will be identified to guide the development of new prompt CWA defeat techniques.

2.Thermal decomposition of DIMP

2.1.Experimental techniques

Various experimental[14,27,28,32,33]and theoretical[29,34]studies have addressed rates and mechanisms of thermal decomposition of DIMP and other sarin surrogates.Among the experimental methods found in the literature,the following can be distinguished:a tubular flow reactor[14,15],batch reactor[27,33],shock tube[32,35-37],and a flow reactor involving plasma[38,39].

Gas phase pyrolysis of DIMP in a tubular flow reactor was studied in Ref.[14].DIMP decomposition was studied in atmospheric pressure nitrogen at temperatures of 700-800 K with residence times of 20-80 ms.A simplified diagram of the experimental setup and the respective DIMP temperature profile are schematically shown in Fig.3.Nitrogen,serving as the main carrier gas,is preheated to the target reactor temperature.Liquid DIMP is first injected into a separate nitrogen flow preheated to 120°C(393 K).This 120°C mixture of gasified DIMP and nitrogen is injected into the main carrier gas.Upon injection,DIMP rapidly reaches the reactor temperature,although this heating is not controlled and any simultaneous reactions are not quantified.The reactor is maintained at a constant temperature,and gas probes are placed in different locations to test reaction products after different exposure times.Upon entering a sampling port,the product gases were diluted with excess nitrogen to quench reaction if possible.

Product gases are directed to the gas cell of an on-line Fourier Transform Infra-Red spectroscopy(FTIR)analyzer or to liquid nitrogen traps to be collected for Gas Chromatography/Mass Spectroscopy(GC/MS)analysis[14,15].

Fig.3.A simplified schematic of a continuous flow reactor for DIMP pyrolysis[14]and illustrated temperature profile in the reactor.

Generally,the flow reactor is a convenient,versatile,and easy to handle system to study combustion/pyrolysis and CWA decomposition mechanisms.However,to interpret results,it must be assumed that no reaction occurred during pre-heating and,in particular,during the rapid temperature ramp following injection of DIMP into the hot nitrogen flow.Unfortunately,the time of such pre-heating can be comparable to the target exposure times.The effect of this pre-heating process becomes greater for reactors operating at higher temperatures.This fundamentally limits the temperature range for such a reactor.Additionally,reaction during sampling cannot be completely ruled out.The effect of sampling lines can be reduced if they are maintained at a temperature that is sufficiently low for the products to remain stable,while being higher than needed for the formed species to condense.

Rapid heating of liquid DIMP in batch reactor configurations was used recently to study decomposition under dynamic conditions.A pyroprobe[33]and laser ablation[27]were used.In both cases,a substrate with DIMP either on the surface or in pores is placed in an enclosed experimental chamber and is rapidly heated.Fig.4 shows the main components of such experimental setups and a diagram illustrating the temperature change of the sample substrate.The liquid sample is rapidly heated past the boiling point and evaporated.The final temperature of the substrate can be maintained,or the power can be removed causing it to cool.The temperature in the gas phase in the experimental chamber may differ from that of the substrate.

The pyroprobe[33]can be a heating coil or a platinum(Pt)ribbon element.DIMP is placed either in a quartz boat located at the center of the coil or directly on the Pt ribbon.In Ref.[33],the pyroprobe was heated at 10±0.025,1,800±200,and 18,000±1000 K/s to temperatures of 500,1000 and 1200°C(773,1273,and 1473 K,respectively)in nitrogen,air,and in an oxygenrich(31.6%)environment.The final temperature of the pyroprobe was then maintained.The products were continuously analyzed by FTIR[33].Unfortunately,the platinum heater can be a catalyst for DIMP decomposition and affect the decomposition mechanism.Because DIMP is initially in the liquid phase,significant temperature gradients are expected in both liquid and gas phases in the experimental chamber.It is even not clear whether all or most of DIMP evaporated before significant portion of the evaporated DIMP decomposed.In other words,quantifying the reaction rate remains difficult,although some qualitative features of the DIMP behavior at high heating rate were identified.

Fig.4.Schematic diagrams for batch reactor experiments with CWAS placed on a substrate,which is rapidly heated[27,33]and for a temperature profile of the heated substrate.

In the laser heating experiments[27],DIMP is initially absorbed into a porous graphite substrate.A laser rapidly heats the graphite surface with an estimated rate of 1011K/s for the area directly exposed to the beam.The decomposition reaction was studied in a closed chamber in both,atmospheric and oxygen depleted environments.The graphite substrate was heated to temperatures of 1440,2140 and 2830±110 K by varying the laser power.The decomposition products were analyzed by in-situ high resolution mass spectrometer,MS,and ex situ FTIR[27].Despite using the insitu MS,the time-resolved product characterization remains problematic because significant temperature and composition gradients must have developed in the experimental chamber at the employed very high rates for the localized substrate heating.Because only the graphite surface was heated,much lower temperatures are expected in the gas and especially near the chamber walls.The lower temperature of the chamber and walls can cause part of the gas species to condense.Thus,the results of the MS analysis associated with a given substrate temperature may not represent the actual product composition corresponding to that temperature.

In the third experimental approach,the rapid decomposition mechanism of CWAS is studied using shock heating[32,35-37].The experimental set-up as well as the respective temperature profile are shown schematically in Fig.5.The driver section is brought to a high pressure and the diaphragm is ruptured.An incident shock propagates through the driven section and reflects from the end wall.Reactions initiated by the temperature increase due to the incident and reflected shocks are usually tracked optically.Even though the decomposition studies summarized here were on the sarin surrogates DEMP[35]and TEP[34],this system is also expected to be useful for studying rapid decomposition of DIMP.Such rapid decomposition of DIMP generating flammable hydrocarbons was recently shown to shorten significantly ignition delays of shock-initiated reactive gas mixtures containing DIMP[37].

Fig.5.Schematic representation experimental setup for shock heating based on Ref[36].

Fig.6.Experimental setup for corona discharge reactor based on Ref.[38,39].

CWAS are introduced in the driven section of the shock tube;the products can be analyzed optically,e.g.,by laser absorption spectroscopy or registering chemiluminescent emission near the end wall.Thus,the gas phase CWAS is heated by the incident and reflected shocks in two closely spaced steps.In the reported experiments,the reaction atmospheres were air and an oxygen/argon mixture,and the temperature could reach 2000 K[32,35,36].CO,one of the final and most stable products of the reaction involving oxidation of the produced species,was probed experimentally[36].In experiments[37]chemiluminescence emission of the excitedstate hydroxyl radical(OH*)was monitored.Observing one of the forming species without resolving other intermediate decomposition products on a similar time scale is not sufficient to establish or validate the reaction mechanism,however.Shock tube experiments have another limitation:it is exceedingly difficult to probe reactions at relatively low temperatures(e.g.,below 1200 K)with reagents that react relatively slowly,as CWAS.The test time of a typical shock tube is 1-2 ms and although low temperatures can be achieved,the reaction rate for decomposing or oxidizing CWAS becomes too low for the reaction to be clearly detected within the test time.Extending the test time is possible by using a driven extension and employing various driver gas mixtures;however,interpretations of such experiments are more difficult.

In the fourth experimental setup illustrated in Fig.6,the decomposition of CWAS occurred in a corona discharge reactor[39].CWAS decomposed without catalytic reactions as during thermal decomposition,and on similar time scales.Because combustion products of many reactive materials are readily ionized[40],it is suggested that such corona discharge experiments and results are of interest and relevant here for analyzing prompt agent defeat reactions and mechanisms.As shown in Fig.6,DIMP was decomposed in a tubular flow reactor containing a coaxial wire serving as a high-voltage electrode.The reactor was kept at 340 K and at atmospheric pressure.A dielectric barrier corona discharge was produced,and DIMP was fed by both nitrogen and helium flows.It is unclear how DIMP condensation on the reactor walls was prevented.Presence of liquid DIMP could bias the reported analysis of the decomposition products.The concentrations of DIMP and its intermediate and final decomposition products were obtained using molecular beam mass spectrometry and GC/MS.The ionization caused by the corona discharge was observed to accelerate DIMP decomposition significantly.In a recent relevant work[41]sarin was decomposed in a dielectric-barrier discharge(DBD).Similar to corona discharge,a low-temperature plasma was produced in an apparatus including a sarin-carrying gas(air)flown between copper electrodes with one of the electrodes covered with glass.AC voltage was applied to the copper electrodes to generate plasma.The decomposition products were analyzed by GCMS.

Note that the plasmas produced by corona and DBD may not be similar to that generated during combustion of reactive materials.Therefore,further work characterizing the decomposition of CWAS in general and DIMP,in particular,assisted by combustiongenerated plasma may be of significant interest.

2.2.Experimental results

In this part,the types of measurements reported in the literature are reviewed,although the specific reported results are not covered comprehensively.The interested reader is referred to the originalpublications for further details.In different experiments introduced in Section 2.1,different intermediate and final products were detected and quantified.The products found to be most important during DIMP decomposition,their respective abbreviations and molecular structures are shown in Table 1 to streamline further discussion.

Table 1Main products observed in different DIMP decomposition experiments.

Studying pyrolysis of DIMP in the continuous flow reactor(cf.Fig.3)mole fractions of DIMP,propene and IPA with respect to residence time were measured[14].These results are illustrated in Fig.7.As DIMP decomposes,both propene and IPA concentrations increase monotonically.Propene production occurs faster than that of IPA.The increased temperatures accelerate the reaction;at the highest temperature tested,799 K,IPA production is saturated at 35 ms,and remains essentially unchanged at longer exposure times.A similar saturation pattern is also observed for propene.For both propene and IPA,saturation at longer exposure times is more pronounced at higher temperatures.

The data collected at different temperatures were used to construct an Arrhenius expression for the rate constant of DIMP decomposition[14].The decomposition was treated as a unimolecular reaction with the rate constant:

Fig.7.Mole fractions of DIMP,propene,and IPA versus residence time at different temperatures[14].The sequence of temperatures is the same in all three plots.

This reaction rate was used in the computations in Ref.[29],shown in Fig.2 above.

In more recent experiments[33],illustrated in Fig.4,the decomposition of DIMP at high heating rates was studied using a pyroprobe.The concentrations of DIMP,IMP,MPA,IPA and propene were characterized semi-quantitatively,using absorbance at the species-specific wavelengths(without directly calibrating the absorbance to the species concentration).The absorbance data were not giving the actual concentration of the DIMP and the products since it was not calibrated.Results of a typical,timeresolved product analysis obtained processing FTIR peaks are presented in Fig.8.In the experiments shown,the temperature was ramped to 1000°C(1273 K)at different rates and then held constant.To guide discussion,and heating nonlinearities notwithstanding(see supplementary material in Ref.[33]),the dynamic heating section and the nominally isothermal section are distinguished in Fig.8.The heating part is shown as a function of temperature,calculated from the reported heating rates,while the static part is shown as a function of time.

In the experiment performed at the lowest heating rate,10 K/s,the DIMP absorbance initially increases and then begins to diminish during the temperature ramp.The first break in the DIMP signal occurs just above 393 K(120°C),closely followed by the onset of the products IMP/MPA and propene near 543 K(270°C).Continued increase of DIMP up to about 793 K(520°C)suggests continued gradual evaporation from the support or re-evaporation of DIMP previously condensed in colder parts of the chamber.As the DIMP absorbance starts to decrease above 793 K,IMP/MPA decreases as well,although propene shows little change.IPA is barely detected during the entire experiment.

Fig.8.FTIR absorbance of DIMP,IMP/MPA,propene and IPA measured in Ref.[33]when DIMP was heated to 1000 °C by pyroprobe at different rates.DIMP is not shown at the higher heating rates because its signal far exceeds the signals of the products.

For higher heating rates(1,800 K/s and 18,000 K/s),absorbance for propene and IPA increases rapidly only after the temperature ramp is over.It is worth emphasizing that the temperature history reflected by those rates describes heating of the substrate.In both cases,products are actively being generated for about 0.8 s before they decompose themselves(propene)or condense(IPA,IMP/MPA).The IPA absorbance becomes stronger at higher heating rates.At 18,000 K/s,IPA absorbance is the strongest,while,as noted above,it was undetectable at 10 K/s.Conversely,at 1,800 K/s,propene is dominant.Thus,different DIMP decomposition dynamics are certainly observed in these experiments as a function of the heating rate.

The maxima in the signals are determined by the balance between formation of the products and their respective decomposition or condensation.Nevertheless,it may be interesting to note that for both,1,800 K/s and 18,000 K/s,maximum concentrations occur at about 2.5 times the duration of the nominal heating ramp.

Interpretation of the measured time-resolved absorbance curves is difficult because evaporation and decomposition of DIMP must have occurred simultaneously,with different processes prevailing in different locations within the experimental chamber.Strong temperature gradients between the pyroprobe heating element and chamber walls must have led to poor homogeneity of the products.Further,condensation of DIMP and its decomposition products could have occurred at the windows and walls of the chamber.

In another DIMP decomposition experiment aimed to achieve shorter exposure times and higher temperatures,8-ns laser pulses heated a porous graphite substrate with DIMP adsorbed to it[27].This ballistic heating approach resulted in nominal heating rates of 1011K/s for the substrate.The unreacted DIMP and its decomposition products were detected in the gas phase above the substrate.Generally,the observed decomposition products are consistent with those previously observed for thermal decomposition of DIMP[14,31].However,in these experiments performed at very low partial pressures of oxygen,ethylene,acetylene,and acetone were also observed.The results are illustrated in Figs.9 and 10.Not shown in Fig.9 is the measured propene concentration,which does not change appreciably.The concentrations of other products are thus shown in respect to propene.In Fig.9,relative product concentrations are plotted as a function of the oxygen partial pressure;the results correspond to the constant final temperature of the substrate of 2830 K.However,because of strong temperature gradients developing in this experiment,this temperature may not be used directly to compare the specific product formation reported in Fig.9 with other temperature-resolved decomposition product measurements.As expected,concentrations of ethylene/acetylene diminish at greater oxygen concentration,while that of IPA increases.

Fig.9.Effect of partial pressure of oxygen on relative product concentrations with respect to propene at the temperature of 2830 K[27].

Fig.10.Effect of surface temperature on relative product concentrations with respect to propene at the partial pressure of oxygen of 10-6 torr[27].

The effect of surface temperature on the measured species concentrations in the laser heating experiments is illustrated in Fig.10.In this case,the partial oxygen pressure was fixed to 10-6torr.Higher temperatures lead to greater concentrations of small molecules ethylene/acetylene as well as of IPA.Here,as for results shown in Fig.9,no effect on the propene concentration was detected and the relative product concentrations are shown again with respect to that of propene.Note that changes in concentrations of IPA and acetone are comparable to the reported error bars,suggesting that higher accuracy measurements are needed for further interpretations of these observations.

As well as in the pyroprobe experiments,here liquid DIMP was absorbed on a heated surface,making it difficult to determine whether DIMP started decomposing even before evaporation.Note that the surface temperatures shown in Fig.10 were estimated using empirical correlations.This could cause a systematic error in assessing the surface temperature.As in the pyroprobe experiments,strong temperature gradients between the heated substrate and the chamber walls lead to additional difficulties in interpreting the observations.

As noted above,no shock tube experiments(Fig.5)with DIMP decomposition could be found in the literature.Results have been reported,however,focusing on shock-induced ignition delays of DIMP-containing flammable gases[37]and decomposition of different sarin surrogates,e.g.TEP,in both oxygen-free and oxidizing environments[34-36].Here,a brief overview of such experiments with TEP is given following Refs.[34,36]to highlight the utility of this experimental approach.The measured concentrations of CO,the final product of TEP pyrolysis(in inert environment)and oxidation are shown in Fig.11a and b,respectively following Ref.[36].The experimental data are compared with four different theoretical predictions,referred to in Fig.11 as LLNL and updates 1,2 and 3.First,LLNL model accounted for a detailed incineration mechanism for organophosphorus compounds[29,42].TEP decomposition pathway has seven unimolecular elimination reactions,hydrocarbon chemistry and phosphorus reactions.In inert environment(Fig.11 a),the predictions are significantly below the experimental CO concentrations.In case of oxidation(Fig.11 b),the predicted increase in the CO concentration is delayed well after the experimental peak is observed.

Fig.11.Experimental and predicted CO evaluation during TEP decomposition in shock tube a)pyrolysis and b)oxidation[34,36].

The model update 1 reported in Ref.[36]included a modified hydrocarbon chemistry model containing reaction mechanisms of C0-C2with oxygenated fuels over a wide range of experimental conditions from AramcoMech.2.0[43]and phosphorus chemistry from LLNL[29].The concentration of CO predicted in pyrolysis by the update 1 model increased;however,remaining well below the experimental.Similarly,a delay in the CO peak formation predicted for the oxidizing environment became smaller,while remaining much greater than observed experimentally.The model update 2 included modified thermochemistry for the phosphoruscontaining species.New thermochemical parameters were obtained using the group additivity method and from quantum chemistry calculations(CBS-QB3 level of theory).However,no significant improvement in the predictions was achieved by update 2 as seen in Fig.11.

Finally,in a recent follow-up study[34],the kinetic mechanism for TEP decomposition was further improved by adding another(eight)reaction for the phosphorus-bearing species as well as an alternative TEP decomposition mechanism.The added decomposition mechanism involves H-abstraction,radical decomposition,and recombination reactions.Also the reaction rates of eight phosphorus-containing elimination reactions were updated using conventional transition state theory based on CBS-QB3 quantum chemistry calculations using Gaussian 09[44].The results are shown in Fig.11 as update 3.For both pyrolysis and oxidation of TEP,the predictions are in better agreement with the experiments.Concentration of CO in inert environment is described approximately.The delay between the experimental and predicted CO concentration peaks in the oxidizing environment is diminished.The improvement achieved by the model update 3 may be because of an increased sensitivity of the predicted CO yield to H-abstraction reaction omitted previously.Although H-abstraction is slower than elimination reactions[29],calculations suggest that in oxygenated environment it might have a pronounced effect.Despite an improvement,it appears that further refinement of the model will be necessary to predict accurately formation of CO as a result of pyrolysis and oxidation of TEP.

Reaction model modifications are also needed based on comparison of the measured and predicted ignition delays for shockinitiated DIMP-containing reactive gas mixtures[37].The calculations used the same LLNL model[29,42]as well as its updates similar to the updates 1 and 2 described above.While the models described well ignition delays for the neat gas mixtures(with no DIMP);the reduced delays caused by added DIMP could not be predicted.Even more alarming,the discrepancy between the measurements and experiments increased for the updated model calculations.

Species detected during decomposition of DIMP carried by nitrogen through a plasma produced by a corona discharge(Fig.6)are shown in Fig.12[39].The decomposition products were normalized by the inlet mole fraction of DIMP and are shown with respect to the specific deposited energy,defined asEX=PF-1X0-1,wherePis the consumed power,Fis the gas flow rate andX0is the initial DIMP concentration.Details reported in Ref.[39]were insufficient to recast the data in terms of exposure time for more direct comparisons with thermal decomposition experiments.Nevertheless,data show that primary intermediate products of DIMP decomposition are IMP and MPA,which is consistent with the products of thermal DIMP decomposition.Also,a new decomposition product orthophosphoric acid(OPA)was observed.Unfortunately,the specific mechanism producing OPA was not discussed.The ionized species formed in the corona-generated plasma(and,possibly,thermal plasma)may attack a broader range of molecular bonds and cause formation of different intermediate decomposition products compared to those forming during purely thermal decomposition of DIMP and similar organophosphorus compounds.It is also interesting that OPA was also observed in the products of sarin decomposition in DBD[41]further highlighting the significance of new reaction pathways associated with ionized species.Sarin was decomposed to isopropyl methyl phosphonic acid(IMPA)and further to MPA and OPA.It was suggested that the break of the phosphorus-carbon bond of MPA might have been caused by its interaction with H+ions and OH radicals leading to the formation of OPA[41].

Fig.12.Mole fractions of DIMP and decomposition products with respect to specific energy,(P is the consumed power,F is the flow rate and X0 is the initial concentration)[39].

Because experiments were conducted at low temperature,it becomes clear that gas ionization accelerates DIMP decomposition.Combustion products of reactive materials are expected to contain ions and generate plasma[40];thus,the effect of ionization on the DIMP decomposition rate might need to be further analyzed to understand prompt CWA defeat mechanisms.

3.Comparison of different experimental techniques used to study thermal decomposition of CWAS

Table 2 lists the experimental systems and respective experimental conditions used in the literature to better understand the thermal decomposition mechanism of DIMP and some other sarin surrogates.The advantages and disadvantages of the reported experiments are summarized.All techniques discussed here are versatile and can be used with a variety of CWAS,even though the results reported to date are limited,in most cases to DIMP,or other selected CWAS.

When selecting or designing a specific experiment,it is useful to consider the temperature and exposure time ranges explored and available for future exploration,which can be associated with the experimental methods discussed here.Graphically,such ranges are approximately illustrated in Fig.13.For convenient reference,temperatures and corresponding exposure times for DIMP reduction from 90 to 99.999%calculated from results reported by Glaude et al.[29],are shown.

Tubular flow reactors,representing the earliest method to study thermal decomposition of DIMP cover a relatively narrow range of times and temperatures[14].Because the reactor is isothermal and sampling ports are spatially separated,the range of temperatures and exposure times for this method,as shown in Fig.13,is well defined(which,as discussed below,is not the case for other techniques.)The accuracy of the product concentration measurements can be rather high because the analysis time is effectively unlimited.Adjusting the tube diameter and carrier gas flowrate enables one to tune the residence time with a high accuracy.However,achieving higher temperatures or shorter times is problematic.Moving to higher temperatures involves pre-heating DIMP before it enters the isothermal reactor;such pre-heating leads to a poorly quantifiable initial DIMP concentration with errors growing exponentially with the reactor temperature.The shorter exposure times are limited by distance between sampling ports and a flow rate suitable for the experimental setup.The range of times and temperatures shown in Fig.13 and reported for the tubular flow reactor does not overlap with the predicted conditions of interest leading to significant decomposition of DIMP.Thus,the reaction kinetics implied by the tubular flow reactor measurements must be validated by measurements at higher temperatures and/or shorter exposure times.

Shock tube[32,34-37]is another established method for probing the rapid decomposition reactions for CWAS.It generates a stepwise heating and enables optical observations of the decomposing species.Characteristic time of the temperature increase is less than 0.1 ms;the formation of optically active products is registered in the ms time range.In Refs.[36],adsorption by CO was measured during several ms in oxygen-free environment,and in less than 0.1 ms in presence of oxygen(Fig.11).Respectively,inFig.13,we limit the time range as 0.05-3 ms.The upper time limit is not well defined because reactions can occur and be detected at later times.The temperatures are readily tunable as long as they are above 1000°C;the gas environment can also be adjusted.These conditions match the high-temperature range of interest for rapid thermal decomposition of DIMP as predicted by the present model(Fig.13).However,the diagnostics are limited to optical(most commonly,absorption)spectroscopy and only few abundant species,typically the decomposition end products,e.g.,CO[36]can be detected.Thus,detecting rapidly forming intermediates or species produced at low concentrations is a challenge.

Table 2Experimental techniques used for thermal decomposition of DIMP and other CWAS.

Fig.13.Characteristic temperatures and exposure time achievable for experimental studies of thermal decomposition of DIMP using different techniques.

To consider dynamic decomposition processes,temperature jump experiments were introduced recently.Either laser beam[27,28,31]or pyroprobe heating[33]were used to observe how a rapidly heated CWAS decomposes.In such experiments,DIMP was adsorbed to or placed on a solid substrate;the substrate was heated.Depending on the heating source,heating rates could vary readily.

A relatively wide range of heating rates and thus implied exposure times was achieved in the pyroprobe heating[33]experiments;this range is schematically shown in Fig.13.It covers relatively low temperatures and extended exposure times matching well with the ranges of interest based on comparison with the calculations.The lower limit for exposure times for this experiment was estimated based on the reported time-resolved measurements.A significant shortcoming of these tests is that the heating rates were defined for the substrate,not for the adsorbed or evaporated DIMP.Strong temperature gradients in the reaction chamber make it difficult to interpret the measured concentrations of decomposition products sampled from the chamber.The concentration measurements in the pyroprobe heating experiments were performed in real time;thus,the signals were less reliable than those obtained from tubular reactor experiments.

In the laser heating experiments,temperatures were variable in a broad range,including very high temperatures.The duration of the laser pulse,8 ns[27],was effectively the heating time for the substrate.However,it must have taken longer to heat,evaporate,and decompose DIMP adsorbed onto the substrate.No timeresolved measurements were reported for these rapid heating experiments,although it was assumed that most of the reaction products formed as a result of the pulsed heating of DIMP on the substrate.Thus,exposure times for the laser heating experiments in Fig.13 are shown very approximately,increasing up to about 1 ms,which might be a reasonable guess for the time it took to heat the absorbed CWAS in Ref.[27].Pulses repeated at 20 Hz for 2 min were used to generate detectable concentrations of the products leading to a slow upward drift in the initial substrate temperature.This added to the uncertainty of the measurement.Note that longer laser heating pulses yielding effectively lower substrate heating rates may be readily achieved,although no such result has been reported.Lack of time-resolved measurements and significant temperature gradients in the experimental setup are main deficiencies of this approach.

From Fig.13,it is apparent that the exposure time range of 10-100 ms has not been resolved at all in the reported experiments.The range of low temperatures and long exposure times was addressed;however,as noted above,the uncertainty in the actual temperature of the decomposing CWAS makes the result difficult to interpret quantitatively.Finally,a much better resolution of the intermediate products is desired for shorter exposure times and high temperatures,which is where future experimental efforts should also be directed.

A tubular reactor combined with a corona-generated plasma is of interest as initially quantifying the effect of ions on the decomposition(not represented in Fig.13 because of the lack of heating).The corona plasma needs to be further characterized and compared to that produced,for example,by burning metallized explosives.In addition,the effects of thermal and plasma-induced decomposition must be clearly separated from each other,and effects of nonhomogeneity of the CWAS-laden gas flow passing through the reactor must be addressed.

4.Pathways of DIMP decomposition

The mechanisms for thermal decomposition of DIMP in inert environments proposed in the literature are shown schematically Fig.14.The mechanism proposed by Zegers et al.[14],is based on isothermal measurements in a tubular flow reactor.It is applicable to the low-heating rate conditions and includes two stages.During the first stage,a hydrogen atom is transferred from one of theβ carbons to the double bonded oxygen through a six-membered ring transition state structure.Next,a propene molecule is ejected and an OH group is formed,while the oxygen that used to connect the propyl group to the phosphorous is now forming a double bond with the phosphorous atom.The resulting phosphorus-containing molecule is IMP.The second stage consists of two competing routes.In the first route,propene is ejected,similar to the first stage,yielding a phosphorus-containing MPA molecule.In the second route,hydrogen from an OH group bonded to phosphorus is moved to alkoxy group to form IPA,while the phosphorus containing MOPO is formed.For the second stage,the propene production route is the favored,based on the results shown in Fig.7,where propene production exceeds that of IPA.

The above mechanism is consistent with the measurements at the lowest heating rate of 10 K/s shown in Fig.8,when the products are dominated by IMP and propene,while IPA concentrations are low.However,IPA becomes more prominent at 1800 K/s at the expense of IMP,which cannot be explained by Zegers’mechanism.It is interesting that IMP/MPA starts to appear just as the target temperature is reached while,at the same time,there is a break in the trend for IPA.This may suggest that the reaction partially reverts to Zegers’mechanism as the environment becomes less dynamic.Finally,IPA dominates throughout at 18,000 K/s,again inconsistent with Zegers’mechanism.Therefore,Yuan et al.[33]proposed an alternative decomposition mechanism for high heating rates and high decomposition temperatures.That modified mechanism,labeled in Fig.14 as relevant for high heating rate,includes only one stage.It was proposed that a different sixmembered ring formed before DIMP transferred hydrogen between the two alkoxy OCH(CH3)2groups and simultaneously formed propene,MOPO and IPA.For low heating rates(10 K/s),when no IPA was observed,only one pathway forming IMP and propene was suggested with further IMP decomposition to MPA and propene,essentially following Zegers’mechanism for low heating rate,with only route 1 being operative.Clearly,in these experiments and given the inhomogeneous temperature distributions,all suggested reaction pathways occur at the same time.Measurements with better control over the environment and over the DIMP supply,as well as with better quantification of the products are needed in order to determine rates for the suggested reactions.

Fig.14.DIMP decomposition mechanisms in inert environments proposed in the literature[14,33].

The effect of heating rate on the decomposition mechanism of DIMP can be understood considering that different reaction stages and routes may have different activation energies.At low temperatures,low-activation energy reactions begin,generating products,necessary for the subsequent steps of decomposition.However,if the temperature is raised quickly,there may not be enough such products formed before the higher activation energy reactions become possible.Indeed,Eilers at al[33].estimated the activation energies for both high and low heating rate pathways computationally with density functional theory(DFT).It has been found that the activation energies of the high and low heating rate pathways are 51.2 and 36.5 kcal/mol,respectively.The latter value is consistent with that reported in Ref.[14],36.7±4.9 kcal/mol for the lowheating rate reaction pathway.Thus,understanding and characterizing kinetics for different decomposition steps is necessary in order to develop a comprehensive mechanism sensitive to the heating rate.

Finally,results shown in Figs.9 and 10 suggest that high temperatures lead to formation of multiple smaller chain products,e.g.,acetone,ethylene/acetylene as a result of further fragmentation occurring after DIMP decomposition into IPA and propene.These additional products may further affect the reaction mechanism,especially,at high heating rates.

It was also observed that in oxygen-deprived environments,the concentration of oxygenated products decreases sharply relative to alkene and alkane(propene and/or ethylene/acetylene)products suggesting that oxygen may be directly incorporated into the products of DIMP fragmentation leading to a different mechanism[27].The alkoxy radical abstracted from DIMP can react with atmospheric oxygen to form alkyl radicals.Alkyl radicals then decompose into IPA and acetone[45].This can further affect the decomposition/incineration mechanisms and rates.These additional reactions should certainly be included in the mechanisms of DIMP decomposition occurring in air or other oxygenated environments.

5.Conclusions

Thermal decomposition and incineration occur in most cases when CWA or their simulants,CWAS,are destroyed.Thermal decomposition is a particularly relevant pathway for destruction of CWAS in prompt agent defeat scenarios.Thus,mechanisms of thermal decomposition of organophosphorus compounds,contained in sarin and its surrogates need to be understood and described quantitatively.DIMP is the CWAS best suited to represent sarin’s behavior in laboratory experiments,from which reaction mechanisms and kinetics can be recovered.

Experiments characterizing thermal decomposition of DIMP were performed using a tubular flow reactor.Propene,IPA,MOPO,and MPA were the most significant decomposition products.That work yielded the DIMP decomposition rate,which was described as a unimolecular decomposition.This reaction rate was incorporated in a comprehensive model,accounting for fluorine chemistry,in particular kinetic data for P-F bonds.This model was extrapolated to higher temperatures and shorter times.However,the decomposition products obtained at high temperatures,i.e.,approaching 1273 K were not characterized experimentally and thus theoretical predictions require further validation.Indeed,products observed in the pyroprobe heating experiments,involving higher temperatures and dynamic environments,were different from those identified in the tubular flow reaction measurements.Moreover,in oxygenated environment a different DIMP decomposition pathway can be possible which directly produces IPA and acetone.Unfortunately,the robust tubular flow reactor approach cannot be readily adopted for high-temperature measurements because the decomposition,which starts occurring in the preheating section of the flow system,is not detected,or accounted for.

The effect of heating rate on DIMP decomposition was addressed experimentally using experimental configurations involving DIMP coated on or adsorbed into a substrate.The substrate was heated,either electrically or by a laser beam,reaching high temperatures quickly.The results show that no IPA formed at relatively low heating rates.The sequence of the formation of various decomposition products and relative amounts of the products formed were affected by the experimental heating rate.Additional products,including smaller hydrocarbons,were found to form,especially at elevated temperatures and in oxygen-depleted environments.However,quantification of the reaction products in such experiments is difficult.Significant temperature gradients in the sealed chamber,in which DIMP is heated and evaporated lead to uncertainty of the analyte.For measurements performed in real time,the sensitivity and resolution become problematic.

In related efforts,rapid decomposition of a different CWAS,TEP and other sarin surrogates,were investigated using shock tube experiments.Ignition delays measured for shock-initiated reactive gas mixtures containing DIMP suggest a strong effect of hydrocarbons forming rapidly upon DIMP thermal decomposition.Although no measurements reported temporally resolved intermediates formed from such decomposition of DIMP,it was shown that shock tube measurements are useful in clarifying high rate decomposition processes for organophosphorus compounds.

Decomposition of DIMP prompted by its interaction with a plasma produced by a corona discharge was also described in the literature.Rapid decomposition occurred at low temperatures,suggesting a substantial acceleration of decomposition reactions.Further,decomposition products were distinct from those observed in DIMP decomposition at elevated temperatures,e.g.,including OPA,further pointing out at alternative reaction pathways.

Future work is desired to quantify rapid reactions occurring at a broad range of temperatures,which would allow one to develop a validated model for thermal decomposition of DIMP and other CWA surrogates.In particular,exposure times of 10-100 ms are of interest as well as species-resolved measurements for the intermediate products formed at high temperatures and even shorter exposure times.Further,it is of interest to consider synergistic effects caused by thermal decomposition,oxidation,and catalytic reactions,which might occur in practical situations.Finally,the effect of ionized species in the environment may need to be further clarified as accelerating or altering the decomposition reaction pathways.

Acknowledgment

This work was supported by the US Defense Threat Reduction Agency,Award HDTRA1-19-1-0023.