Shomik Mukhopadhyay,Mirko Schoenitz,Edward L.Dreizin

New Jersey Institute of Technology,Newark,NJ,USA

Keywords: Chemical weapon agents Prompt defeat Metal combustion Heterogeneous reactions

ABSTRACT Chemical warfare agents(CWA)are stockpiled in large quantities across the globe.Agents stored in inaccessible facilities need to be destroyed rapidly without dispersing the compounds to surrounding areas.Metal-based energetic formulations are used in such prompt defeat applications to rapidly decompose the CWA by generating a high temperature environment.An alternate,and possibly a more effective decomposition pathway could be provided by chemicidal action of aerosolized condensed combustion products,which typically consist of metal oxides.Toxic fumes that escape the high temperature blast zone can be neutralized by smoke generated during combustion,depending on the particle size,surface characteristics,chemical properties,and concentration of this smoke.This review considers relevant experimental and modeling studies quantifying decomposition of CWA comprising organophosphorus compounds and their surrogates on the surface of various metal oxides.Dimethyl methylphosphonate(DMMP),a sarine surrogate,was used most commonly for such experiments.Many reported efforts focused on the mechanisms of adsorption of DMMP to various metal oxides and initial reaction steps cleaving various bonds from the chemisorbed molecules.For selected oxides,these experiments were supported by quantum-mechanical calculations.In other studies,the capacity of oxide surfaces to adsorb and decompose DMMP was quanti fied.In most cases,porous catalysts were used although limited experimental data are available for aerosolized nonporous oxide particles.The reported experimental data applicable to scenarios involving prompt decomposition of CWA are summarized.It is noted that information is lacking describing respective heterogeneous reaction kinetics.Preliminary estimates of aerosolized smoke particle concentrations required to destroy CWA are made considering gas phase diffusion rates and reported values of the oxide capacity to decompose CWA or their surrogates.

1.Introduction

Chemical warfare agents(CWA)are toxic synthetic chemicals having rapid,lethal effects on humans that can be dispersed as a gas,liquid or aerosol[1].These compounds have in flicted thousands of casualties[2].CWA can be classi fied by their historical development,physicochemical features,reactivity,or target organs[3].Some commonly known CWA are sarin,VX,chlorine gas,mustard gas,hydrogen cyanide and phosgene[4].Destruction and disposal of CWA are challenging because these compounds need to be altered chemically.Technologies exist for decomposing CWA stored in accessible facilities[5,6].However,there are limited strategies to destroy CWA stockpiles,which are inaccessible to customized equipment.While it is desired to destroy CWA stockpiles promptly,it is necessary to avoid dispersion of CWA or their toxic byproducts to surrounding areas.One available strategy for prompt agent defeat involves incinerating the area using warheads containing conventional explosives(RDX,TNT etc.)[7].

However,conventional explosives produce a pressure pulse,which is estimated to disperse up to 5%of the toxic compounds to surrounding areas[8].Also,energy yield of conventional explosives may not be sufficient to thermally decompose all stored CWA[8].Proposed modi fications for such warheads include augmenting incendiary compositions by adding metal-oxidizer mixtures,thermites[9],and other metal-based compounds(boron,titanium,etc.)combined with such oxidizers as lithium perchlorate,and including wicking compounds to extend the burn times[10].Combustion of these formulations generates high temperatures,at which CWA get pyrolized;they also produce condensed phase metal oxide particles(e.g.,alumina,titania),which can enhance CWA decomposition.

At high temperatures(about 2000 K)developed inside the blast produced by an explosive charge,CWA decompose rapidly by pyrolysis.Conversely,CWA left outside the direct blast volume or dispersed by the pressure pulse may be exposed to lower temperatures and remain intact.At temperatures below 750-800 K,exposure times of several seconds may be needed[11,12].Elevated temperatures may not be sustainable for such relatively long times,however.An accelerated,low-temperature decomposition pathway should thus be made available.One such pathway may be enabled by condensed phase metal oxide particles(smoke)generated during combustion.Metal oxides may react with CWA directly or serve as catalysts for their destruction[13,14].This review is focused on the relevant gas phase reactions that may lead to prompt decomposition of CWA in presence of metal oxides.No additional experiments or computations were performed to prepare this paper.A new interpretation enabling one to predict an upper bound for prompt decomposition of CWA interacting with an aerosol of metal oxide is offered in Discussion section of the article.

Aside from metal oxides,materials reported to adsorb and decompose CWA include metal organic frameworks,activated carbon,enzymes[15,16],hydroxides[17,18],oxyhydroxides[19]and carbon containing formulations[20-24].However,those materials are not components of energetic formulations and are not readily produced as high-temperature combustion products.Substantial previous work focused on the use of metal oxides for decontaminating aqueous solutions of CWA and their simulants[25-31].These studies were conducted at room temperature,with reactions taking place for extended durations.Such reactions and processes will not be considered here.

Presently,no experimental data were found directly describing the real time interaction of CWA vapors with metal oxide smoke produced by combustion of energetic formulations.Instead,research has focused on the use of metal oxides in filtration and decontamination systems.Most of these studies considered decomposing nerve agents,especially sarin[12,32-38].Sarin(isopropyl methylphosphono fluoridate)belongs to the family of organophosphate compounds and is a highly toxic compound[39,40].Toxicity of several CWA,including sarin,as a function of their respective vapor pressures at room temperature is illustrated in Fig.1.Toxic concentrations of these compounds are expressed in comparison to carbon monoxide,which has an acute exposure guideline level-3(AEGL-3)or a lethal inhalation concentration value of 1900 mg/m3for a 10 min exposure.The corresponding value for sarin is 0.38 mg/m3,which is 5000 times that of carbon monoxide.Sarin combines both high toxicity and volatility.Large declared stockpiles of sarin(over 277 tons)are present across the globe[41];there is evidence of undeclared stockpiles and recent use of sarin against civilian populations[42].Thus,studies helping to understand the processes involved with prompt destruction of sarin and its stockpiles are of signi ficant interest.

Fig.1.Acute inhalation toxicity of selected CWA vs.their respective vapor pressure[43-47].

Due to sarin’s high toxicity,laboratory experiments are commonly conducted with its respective chemical agent weapon simulants(CWAS).CWAS have molecular structure and chemical properties similar to those of corresponding CWA.Some of the most common CWAS for sarin are DMMP(dimethyl methylphosphonate),DIMP(diisopropyl methylphosphonate),DFP(diisopropyl fluorophosphate),TEP(triethylphosphate)and DEEP(diethyl ethylphosphate)[48].While chemical functionality of sarin is better reproduced by DIMP and DFP,few studies addressed their interactions with metal oxides[32,49-51].Instead,DMMP has been the most common CWAS for such reactions due to its low toxicity and ease of handling[52,53].The structures and some properties of DMMP and sarin are compared to each other in Table 1.

Table 1 Structure and key properties of sarin and DMMP[48].

Although CWAS are much less toxic than CWA,personal protective equipment and carefully designed experimental facilities are required to handle these compounds.For example,DIMP and DFP are acetylcholinesterase inhibitors,while DMMP is listed in Schedule-2 of the Annex of Chemical Weapons Convention as a precursor for the production of sarin[48].

Toxicity of a sarin molecule can be quickly reduced in presence of reactive surfaces,such as metal oxides.As shown schematically in Fig.2,sarin binds to an active site of an oxide adsorbent via its phosphoryl oxygen.Nucleophilic attack by oxygen atom(on surface hydroxyl groups)on the phosphorus center results in cleaving off its P-F bond[32].This leads to formation of isopropyl methylphosphonate(IMPA)and fluorine,which remain bound to the surface.At elevated temperatures,fluorine can desorb as Hydrogen Fluoride(HF)after protonation.The diagram in Fig.2 is simpli fied;it skips important intermediate steps discussed in detail elsewhere[32,34,36,38].As illustrated in Fig.3,both HF and IMPA are much less toxic than sarin[45,54],however potential release of HF into surrounding areas remains a concern.In the following reactions,loss of an isopropyl group from the IMPA generates methylphosphonic acid,MPA,which is also several orders of magnitude less toxic than sarin[55].

Fig.2.A simpli fied diagram showing initial steps of decomposition of sarin on the surface of a metal oxide[32].When one of the P-O-M bonds breaks,IMPA is produced.

Fig.3.Reduction in acute oral toxicity for products of decomposition of sarin[45,48,54,55].Note that the toxicity value marked for HF represents inhalation toxicity.

In DMMP,as illustrated in Fig.4 following Ref.[51,56],a methoxy group is attached to phosphorus instead of fluorine in sarin molecule.During decomposition,DMMP binds to a metal oxide surface just as sarin,via its phosphoryl oxygen.Nucleophilic attack by oxygen atoms in surface hydroxyl groups on the phosphorus center results in the bidentate structure illustrated in the middle of Fig.4.Instead of the analogous P-F bond in sarin,for DMMP,its methoxy group is first cleaved and protonated to generate methanol[32,51].This leads to formation of(adsorbed)methyl methylphosphonate(MMP).In the following reaction,the second methoxy group is cleaved,analogously to the cleavage of the isopropyl group for sarin.The phosphorus-containing product is methylphosphonate(MP).These reaction similarities make the work on the decomposition of DMMP relevant as providing insight into reactivity of metal oxides towards sarin[12,53].Although several relevant reviews describing the role of metal oxides in decomposing CWA and CWAS,such as DMMP are available[57-60],this paper offers a summary focused on reactions of vapor phase DMMP on surface of metal oxides relevant for prompt defeat applications.

Fig.4.Schematic diagram showing initial steps of decomposition of DMMP on surface of a metal oxide[51,56].

2.Research methods

2.1.Surface based experimental methods

Decomposition of DMMP adsorbed to heated surfaces of metal oxide powders was studiedin-situusing a variety of spectroscopic techniques,including inelastic electron tunneling[51,61],Fourier transform infrared spectroscopy(FTIR)[37,50,56,62-74],x-ray photoelectron spectroscopy(XPS)[65,66,75-83],reflection absorption infrared spectroscopy(RAIRS)[84],and Auger electron spectroscopy(AES)[85].Such studies utilize a type of the experimental setup shown schematically in Fig.5.The substrate(commonly a pressed oxide powder)is heated and evolving species are analyzed.Speci fic temperatures,at which DMMP or its decomposition products desorb from the substrate upon heating are often obtained using temperature-programmed desorption(TPD)[37,66,75,83,85],which can be directly coupled with a spectroscopic detector,e.g.,a mass spectrometer[75,83].

Environments vary in these studies,and can include high vacuum(e.g.Ref.[82]),or low pressures of oxygen(e.g.Ref.[68]).In most experiments,temperatures range from ambient to 673 K(400°C)[51,56,61,69].

Fig.5.Schematic illustration of experimental setup for study of DMMP decomposition on metal oxide surfaces[51,56,61,67,69,84].

Vibrational spectra of functional groups present on molecules adsorbed on metal oxide surface at different temperatures help identifying the structure of decomposition products of DMMP.Through these studies,decomposition products,such as MMP and MP evolved over oxides of alumina[51,61,64]and magnesia[56]have been clearly identi fied.Insights into decomposition mechanism over other oxides,such as iron oxide[56,83,85],cerium oxide[75]and copper oxide[65,78],have also been gained.

2.2.Reactor based experimental methods

To study bulk reactions,batch or flow reactor con figurations have been used.In a batch reactor experiment illustrated in Fig.6a[86],DMMP vapor was injected into an air-filled chamber at room temperature and atmospheric pressure at a concentration representing about one tenth of its saturated vapor pressure.Upon injection,DMMP gets adsorbed on reactor walls.A metal oxide(TiO2,anatase)powder,activated by ultraviolet(UV)light,was also injected into the chamber(once the concentration of DMMP inside the chamber stabilized)and maintained in aerosol using a vibrating diaphragm.The chamber was connected to a sampling system circulating the air through an FTIR spectrometer.Time-resolved measurements for concentrations of DMMP and its decomposition products were recorded.The primary interaction of DMMP vapor and aerosolized particles occurs as aerosol particles are injected.The rate of decontamination was judged from a drop in DMMP concentration with time during this regime.Secondary reactions of DMMP desorbing from reactor walls and interacting with aerosol particles occurs much more gradually.

In a more typical flow reactor experiment illustrated in Fig.6b,DMMP is passed over a fixed bed of a catalyst[12,53,73,87-99].Although most of the relevant studies were designed to select,characterize,and optimize oxides for low-rate processes typical of filtration systems,the results describe decomposition of DMMP vapors of interest to this review.Commonly,a DMMP vapor suspended in a carrier gas(air,helium,or nitrogen)is introduced in a heated tube(reactor)and passed over a metal oxide powder placed inside the tube.The product stream flows into a gas analyzer,e.g.,a gas chromatography(GC)column as shown in Fig.6b.Residence times range from sub-second to several seconds and temperatures in the order of 500-800 K are achieved.Commonly,the plug flow reactor model is applied,and thus,the concentration pro file of the reaction products is not accounted for.

In some studies,the focus is to determine the breakthrough point(BTP)defined as the quantity of DMMP injected into a flow reactor until it starts appearing in the product stream[87,96].In other words,the capacity of the metal oxide in terms of adsorbing DMMP is quanti fied.In some cases,rather than continuous injection,DMMP was injected in pulses[73,100].In selected experiments,instead of passing over a metal oxide surface as shown in Fig.6,DMMP diffused through oxide membranes,e.g.,platinum coated alumina[101-103].The effectiveness of membrane reactors compared to plug-flow systems and effect of temperature and pore size on BTP were studied.While such diffusion processes are rather slow and involve catalytic reactions with noble metals,which are outside the scope for this review,the activity of platinum-coated surfaces can serve as a convenient reference when assessing the performance of conventional metal oxides.Some studies explored use of photosensitive titania surfaces for decomposing a flowing CWA vapor while being activated by a UV light source[36,38,104-107].

Fig.6.(a)Aerosolized titania particles interacting with DMMP vapor[86,108],(b)Schematic illustration of flow reactor containing metal oxide powder[53,73,87-92,96].

2.3.Computational modeling

Many efforts have been made to predict the decomposition of CWA and CWAS using computational models based on Density Functional Theory(DFT)[65,76,79,80,83,109-119].Several types of functionals(such as PBE[65,79,80,83,111,112,114-118]and B3LYP[76,109,110])have been explored to calculate energy states of reactants,products,and intermediate species.Two types of model structures have been considered in these studies:an in finite slab supercell[79,80,83,114,116-118]and cluster[76,109-112,118].In the slab(periodic)models,the unit cell of a metal oxide crystal is repeated in 3-D to represent its bulk structure.The target molecule(such as DMMP or sarin)is introduced on a particular plane(usually the most thermodynamically stable)of this slab.Cluster models consist of a selected fragment of the lattice structure of the metal oxide.A proper design of the cluster is dif ficult;however,this approach may be more versatile and ef ficient.Upon introduction of the target molecule,adsorption enthalpy corresponding to the most stable(least energy)con figuration is recorded.Further decomposition of the adsorbed molecule is based on this con figuration.These models often predict a broad range of possible reactions and intermediate products.DFT calculations are often complemented by results obtained from spectroscopic studies to narrow the range of potential outcomes.The extent of decomposition considered in computational models is commonly limited to the first decomposition step,which is cleavage of one bond from the target molecule.The transition state theory has been utilized in some studies to calculate pre-exponential terms for estimating the rate of decomposition[113,116,117].

3.Results

3.1.Adsorption and decomposition of DMMP on metal oxides

The vibrational spectra of DMMP adsorbing to and decomposing on surface of different metal oxides(cf.Fig.5)have been recorded.The identity of decomposition products was determined by comparing recorded spectral signatures with those of known compounds.To illustrate such studies,vibrational spectra of DMMP adsorbed at room temperature onto Y2O3[84],TiO2[69],CuO[65],and WO3[71]are shown in Fig.7 over the wavenumber range of 1000-1350 cm-1.This low-frequency region was selected to highlight the phosphorus-oxygen bond vibrations.For DMMP vapor,the peak intensity of P=O stretching mode appears at 1276 cm-1.This mode appears at lower frequencies(wavenumbers)for DMMP adsorbed on metal oxides.For example,peak intensity of P=O mode for DMMP bonded to CuO appears at 1242 cm-1.This red shift is caused by strong interaction between the phosphoryl oxygen and an acidic site on metal oxide,which can be either metal(Lewis acid)site or surface hydroxyl(Bronsted acid)site.

Fig.7.Low frequency IR spectrum for DMMP vapor and DMMP adsorbed on CuO[65],WO3[71],TiO2[69]and Y2O3[84]at room temperature.

In the spectral pattern for DMMP on TiO2,two distinct features were identi fied for the phosphoryl oxygen vibrational mode.Peak at 1235 cm-1(marked with+)corresponds to DMMP H-bonded to a surface hydroxyl,while peak at 1215 cm-1(marked by a dotted vertical line)corresponds to DMMP chemisorbed onto a Lewis acid site.

Both con figurations for adsorption of DMMP to metal oxide are shown schematically in Fig.8.The recorded spectral signatures enable one to determine the preferred way DMMP adsorbs to the surfaces of different oxides and how temperature affects the adsorption ef ficiencies.Analyzing spectroscopic data for DMMP bonded to metal oxides at various temperatures,respective decomposition mechanisms are proposed.DMMP molecules adsorbed to the Lewis acid sites decompose upon heating,whereas those physisorbed to hydroxyl groups(Bronsted acid sites)do not,and simply desorb instead[61].The nature of bonding between DMMP and the metal oxide surface has also been studied analyzing hydroxyl stretching modes(not presented here)near 3500 cm-1[69].

Fig.8.Types of bonding sites for DMMP on metal oxide surface:hydrogen bonded to hydroxyl(Bronsted acid)site and chemisorbed onto undercoordinated Lewis acid site.

Another key vibrational mode visible in the low frequency region(1187 cm-1,Fig.7)is that of O-P-O(cf.Fig.4,the intermediate adsorbed complex),which occurs when DMMP decomposes on the oxide surface.This peak is very close to that of O-CH3,but was clearly distinguished for CuO[65],TiO2[69]and Y2O3[84]as shown in Fig.7.The O-P-O mode is visible on Y2O3also at 1092 cm-1.Detection of the O-P-O mode is indicative of the beginning of formation of MMP already at room temperature,at which the spectra shown in Fig.7 were obtained.Concurrently,the peaks appearing at 1108 cm-1for CuO and 1120 cm-1for TiO2surface correspond to the methoxy group bound to a metal atom,which also indicates the decomposition of the adsorbed DMMP occurring at room temperature.

Bond cleavage in the DMMP molecule is further understood by studying vibrational spectra of carbon containing groups,which appear at higher frequencies(2700-3000 cm-1).Methyl and methoxy stretching modes for DMMP on Al2O3and Fe2O3are compared in Fig.9 for the temperature range of 25-400°C.Vibrational spectra are represented in Kubelka-Munk units,where the concentration of a surface bound species is a function of the respective peak area for the absorption intensity pro file[120].The intensity of methoxy stretching modes for DMMP on Al2O3at 200°C at is half of that at 30°C(at 2956 and 2853 cm-1),indicative of a surface-bound MMP species.Further,heating to 200°C leads to disappearance of all methoxy groups,but P-CH3mode is still visible(at 2997 and 2932 cm-1),indicative of the surface-bound MP.The methyl group remains strongly bound to the phosphorus atom across the entire temperature range.On the contrary,intensity of carbon-bonded groups reduces simultaneously for DMMP decomposing on Fe2O3surface.The spectrum remains nearly unchanged up to 100°C.Subsequent heating results in a loss of both methoxy and methyl modes.All carbon containing species are lost fromthe surface at 300°C,leaving behind a phosphate layer[56].

Fig.9.Comparison of methyl stretching modes for DMMP adsorbed on alumina and iron oxide between 30 and 400°C[56].

As illustrated in Fig.4,surface hydroxyl groups serve an essential purpose in the decomposition of chemisorbed DMMP by cleaving and protonation of methoxy groups leading to a bidentate structure(O-P-O).In turn,concentration of H-bonded hydroxyl groups reduces with temperature as they are consumed.Decomposition of DMMP proceeds via either stepwise or simultaneous loss of carbon-containing groups.The stepwise decomposition involves loss of each methoxy group followed by the methyl group to produce MMP and MP in succession with methylphosphonic acid(MPA)as an intermediate.This pathway presented schematically in Fig.10a was proposed for DMMP decomposition over Al2O3[61],a non-reducible oxide.On the other hand,a reducible oxide Fe2O3provides alternate,low-energy decomposition pathways(Fig.10b)where methyl and methoxy groups are cleaved and oxidized by lattice oxygen.Cleavage of both groups occurs almost simultaneously,producing various kinds of surface bound species.Heating the surface leads to loss of carbonaceous products in form of CO,CO2,formic acid and methanol.For DMMP adsorbed to oxides of iron and copper,the double bond between phosphorus and oxygen was found to be retained,leading to phosphate residue.Lattice oxygen becomes available for reaction upon partial reduction of the oxide,e.g.,from Fe3+to Fe2+[56,85].This reaction of surface-bound species with lattice oxygen and desorption of the oxidized products is referred to as the Mars van Krevelen mechanism.In general,depleted lattice atoms can be regenerated and metal oxide surface reactivated upon passing oxygen gas over the surface[121,122].

As stated previously,the formation of MMP is an important step in the DMMP decomposition because it resembles P-F cleavage in a sarin molecule.Both the P-F bond of sarin and the methoxy group of DMMP adsorbed to metal oxides are cleaved at room temperature[32,61].Subsequent decomposition steps involving loss of alkyl and alkoxy groups from sarin lead to a reduced toxicity and generate a phosphate surface layer,which may have catalytic properties.This is similar to the reactions leading to the formation of MP and phosphates shown in Fig.10.Thus,tracking decomposition steps of DMMP(and,likely other CWAS)could be instructive for understanding analogous steps in destruction of sarin.Speci fic temperatures,at which decomposition products of DMMP were observed in different studies,are summarized in Table 2.Decomposition of DMMP on TiO2,CeO2and WO3is reported to be similar to that on Al2O3,involving a stepwise reaction with the DMMP→MMP→MP sequence and retaining the P-CH3bond below 400°C.For DMMP on MgO,minimal decomposition occurs at room temperature,forming MMP at higher temperatures.Formation of MP is not observed even at 500°C[74].The summary in Table 2 shows that for the same oxide,the temperature at which DMMP molecules begin decomposing can vary.For example,a custom alumina surface was generated by subjecting a thin aluminum film to a plasma discharge.Such an oxide layer decomposed DMMP into MMP at room temperature[61].Conversely,formation of MMP on a commercially procuredγ-Al2O3was not observed until 200°C[56].Variation in DMMP decomposition temperatures is also observed for magnesia samples prepared differently.Fabrication and processing techniques such as annealing temperatures affect degree of hydroxylation on the oxide surface and its surface area and thus could in fluence low temperature reactivity[56].

Fig.10.(a)Decomposition of MMP into MP on surface of Al2O3[61](b)Oxidized surface bound species found on surface of Fe2O3 decomposing into phosphates[85].

Table 2 Reported temperatures for formation of MMP and MP during decomposition of DMMP on metal oxides.

DMMP decomposition was reported using other oxides,such as silica[37,63,72,85,123],molybdenum oxides(MoO3[66,76,79]and MoO2[80]),and supported systems,e.g.,Ni/TiO2[82],Cu/TiO2[77]and Fe2O3/Al2O3[62].Interaction with silica was limited to physisorption,most probably due to lack of Lewis acid sites.It is observed that DMMP strongly adsorbs to the surface of stoichiometric MoO3at room temperature,however the majority of the adsorbed molecules do not decompose at higher temperatures and desorb intact.A reduced MoO2surface,however,is capable of decomposing DMMP into MMP.Although decomposition mechanism over Ni/TiO2and Cu/TiO2was not described,these oxides had carbon-containing groups on their surface below 400°C.In general,supports and dopings are found to be stabilizing the reducible metal oxides by charge transfer,thus enhancing their reactivity[70,73,97,124].Although not discussed here,the liberated gases have also been analyzed using spectroscopic techniques(such as mass spectrometry[37,75,81])for a more robust identi fication of the DMMP decomposition products forming at different temperatures.

3.2.Metal oxide adsorption capacity

The flow reactor apparatus illustrated in Fig.6b was used in many studies to quantify the DMMP adsorption capacity for different metal oxides.The capacity is commonly quanti fied using the BTP as a practical parameter.BTP of a metal oxide powder placed in a reactor is found experimentally as the number of moles of DMMP that is injected into the reactor before it is detected in the product stream.It is commonly normalized per unit mass of the oxide.

In a typical measurement,the concentration of DMMP in the product stream is monitored with time at a speci fic constant temperature.Such measurements are illustrated in Fig.11 for a flow of DMMP with air as carrier gas over Al2O3,CuO/Al2O3,Fe2O3/Al2O3,NiO/Al2O3,V2O5/Al2O3[92]and Pt/Al2O3[53].Supported oxides of copper,iron,nickel and vanadium on alumina have 10%of the metal by weight.This will be represented by the “10%” label inserted in front of respective components mentioned below.In Fig.11,the conversion of DMMP is obtained as 100%-XDMMP,where XDMMPis DMMP fraction in the product stream.The adsorption capacity of all materials is very high initially,resulting in no DMMP being detected in the product stream at early times.At longer times,DMMP starts being detected in the products,indicating a reduced conversion.BTP for DMMP is calculated for each material per unit of mass,or BTPm,in terms ofμmol of DMMP per gram of oxide(μmol/g)accounting for the concentration of DMMP in air entering the flow reactor,mass of powder used,and the time the conversion declines from 100%.

Fig.11.DMMP carried in air:conversion pro files for alumina and alumina-supported catalysts at or around 400°C[53,92].

After the BTP,the activity of the oxide continues to drop.In some cases,e.g.,vanadia-or platinum coated alumina(V2O5/Al2O3or Pt/Al2O3,Fig.11)the drop is relatively small so that the high degree of conversion is retained.This can be understood examining the surface of oxide before and after the experiment.For V2O5/Al2O3,measurements show that at the BTP,the speci fic surface area decreased from 248.8 to 4.3 m2/g.At the same time,phosphate species(PO43-)were generated after the exposure to DMMP.These phosphates(also existing as P2O5)are capable of catalytically decomposing DMMP long after active sites of the material have been covered[92,93].

Although a wide range of compounds are detected in the exit stream of a flow reactor,including MP,MPA,P2O5,coke and metal phosphates,reaction kinetics could not be quanti fied before BTP,while DMMP was actively adsorbing to a fresh surface.In some cases,as discussed below,kinetics could be quanti fied following BTP,when a steady state regime is observed as shown in Fig.11 for Pt/Al2O3.However,such reactions are deemed less relevant for scenarios important for prompt CWA defeat.

The BTPmfor DMMP on various metal oxides is shown in Fig.12a as a function of temperature.Note that in many experiments,P2O5was detected on reactor walls,and the effect of walls on BTPmcould not be separated from that of the catalyst.The dotted line marks a capacity of 1 g,or 8064μmol,of DMMP decomposed for every gram of powder.This benchmark is exceeded for only two materials.In general,the BTPmincreases with temperature.This signi fies an accelerated decomposition of DMMP at higher temperatures.In order to be decomposed on the surface,DMMP needs to be adsorbed,while the adsorption ef ficiency of DMMP to the surface decreases with temperature.The combined effect of a reduced adsorption ef ficiency and accelerated decomposition rate explains a minimum for BTPmobserved in Fig.12 for certain oxides,for which both high and low-temperature measurements were reported in the literature.

In Fig.12b,BTPmreported per unit mass is converted to that per unit of surface area for the powders present in the flow reactor,BTPA.Comparing trends shown in Fig.12 a and b enables one to separate the effects of the oxide composition from that of the available surface area.For example,activated carbon shows an outstanding BTPm.When its surface area is considered,its BTPAbecomes comparable to that of alumina.For sol-gel prepared alumina,the BTPmis greater than for a more commonγ-alumina in Fig.12a.However,when corrected for the surface area,γ-alumina shows a greater BTPA,except for one point taken at room temperature.

Fig.12.BTP for DMMP over 1%Pt/Al2O3[89],activated carbon[93],autoclave prepared(AP)CaO[95],copper substituted hydroxyapatite(Cu-HA)[88],sol-gel prepared Al2O3,γ-Al2O3,FeOx/Al2O3[96]and MgO[87]at various temperatures(a)BTPm(μmol/g)(b)BTPA(μmol/m2).

To observe a more direct comparison of BTP for a range of materials,the values reported at the same temperature,400°C,are shown in Fig.13.The inert carrier gas used in these studies is helium.In agreement with Fig.12,activated carbon and 1%Pt/Al2O3demonstrate outstanding BTPm.It is interesting that the oxides combining silica with vanadia and copper oxide with alumina appear quite attractive as well,considering both their mass and surface based capacity values.It was proposed that the resistance of silica towards reaction with phosphates is a contributing factor towards its high capacity[92].Combining iron oxide with alumina also leads to a greater BTPAthan for eitherγ-or sol-gel prepared aluminas.Among considered oxides,copper and iron oxides may be readily augmenting alumina forming as combustion products of respective aluminum-based thermites in oxygenated environments.Thus,testing such thermites as additives to reactive materials generating chemicidal combustion products may be of interest.Data in Fig.13 suggest that BTPmvalues of iron oxide coated alumina[92,96]are higher when air is used as carrier gas.However,when correcting for the surface areas,respective BTPAvalues become similar to each other.The same trend(not shown here)is observed for uncoated alumina used in both studies.This indicates that higher BTPmin Ref.[92]maybe due to larger surface area of oxide powders used.While it has been observed that addition of oxygen during reaction of DMMP with reducible metal oxides reactivates the surface and oxidize surface bound carbonaceous groups[69],this effect may not be re flected in the reported BTP values.Certainly,further work addressing the effect of oxygenated carrier gas on BTP is of interest.

The decomposition of DMMP over both bare alumina and platinum-coated alumina has been extensively studied in the literature.It was reported,as illustrated in Figs.12 and 13,that platinum coating increases BTP for alumina by about 24 times.Similarly,BTP values were compared for sarin,decomposing over bare and platinum-coated alumina in a tubular reactor[12].For sarin,BTP increases by 16 times(at 400°C)for the platinum-coated alumina.Understanding this qualitatively similar increase in the BTP caused by platinum coating for both sarin and DMMP may be instructive for understanding the metal and metal oxide catalytic decomposition mechanisms for CWA and CWAS.

Fig.13.BTP for DMMP over 1%Pt/Al2O3[89],activated carbon[93],10%V2O5/SiO2,10%V2O5/Al2O3,10%CuO/Al2O3,10%Fe2O3/Al2O3[92],copper substituted hydroxyapatite(Cu-HA)[88],sol-gel prepared Al2O3,γ-Al2O3,FeOx/Al2O3[96],MgO[87]and autoclave prepared(AP)CaO[95]at 400°C(a)BTPm(μmol/g)(b)BTPA(μmol/m2).

3.3.Volatile species produced by DMMP decomposing on metal oxides

Volatile species are detected in the flow reactor product stream.Such measurements are illustrated in Fig.14 for DMMP passing over γ-Al2O3.Because of emphasis placed on the time-dependent reactions important for prompt defeat applications,the data from Refs.[96]are processed to express the concentrations as a function of exposure time.In those tests,the flowrate of helium containing DMMP at the concentration of 30.25μmol/L was maintained at 30 mL/min.Individual species detected are shown in Fig.14 a and b for two temperatures,100 and 400°C,respectively.In each case,the dashed vertical lines indicate the BTP.The amount of gaseous species produced rapidly increased with temperature.This is clearly illustrated in Fig.14 c,where the total gaseous species generated by decomposition at different temperatures are shown as a function of time.At low temperatures,decomposition products,such as methanol,remain adsorbed on the surface.Continuing exposure to DMMP leads to saturation of active sites.Some surfacebound species desorb,most likely due to interaction with fresh DMMP molecules,generating oxygenated species such as dimethyl ether.In a recent study,formation of dimethyl ether was attributed to decomposition of surface bound MMP [83].At lower temperatures,it becomes the first species detected in the products before BTP.At higher temperatures,the product stream includes large amounts of oxidized species,such as CO and CO2.Most likely,similar reactions will be of critical importance in reaction scenarios applicable for prompt CWA defeat,when surrounding gas contains oxidizing species.Heat released by forming such species might lead to a temperature increase of the aerosolized oxide particles,which in turn will lead to an accelerated reaction kinetics and an increased BTP.

Fig.14.Composition of product stream for decomposition of DMMP added to helium and flown overγ-Al2O3(a)Composition of product stream at 100°C and(b)Composition of product stream at 400°C[96](c)Total product flow at various temperatures*Note different x and y-axes scales.

As the exposure of the oxide surface to DMMP continues,the active metal sites become covered by the decomposition products,such as MP and phosphates.Their further dealkylation and desorption leads to methanol generation.It is consistently observed that deactivated surfaces tend to generate methanol for extended durations.Although activity is reduced in time,theγ-Al2O3surface continues to decompose DMMP for extended durations[96].Such reactions are unlikely to be of importance for prompt CWA defeat.

Reaction products generated by DMMP decomposition on metal oxides are qualitatively similar to those formed during decomposition of sarin.A speci fic comparison is available for platinumcoated alumina.In an oxidizing environment,a fresh,platinumcoated alumina breaks DMMP down into CO2,H2O and P2O5(or phosphoric acid).At the same time,decomposition of sarin(here:CH3PO2F[C3H7])generates the same compounds as well as HF,see Eq.(1)[12].The reaction is exothermic,with an estimated reaction enthalpy of-2705 kJ/mol(calculated values for heat of formation of sarin and phosphoric acid were used[125]).

3.4.Kinetics of DMMP decomposition over metal oxide surface

In experimental work describing overall kinetic relations that are relevant in the context of rapid CWA destruction,two processes can be distinguished:initial removal of DMMP from the gas phase via its reactive adsorption to the surface of a metal oxide,and further decomposition of DMMP aided by the metal oxide.Typical time resolved observable variables are changes in the gas phase concentration of DMMP and the evolution of decomposition products.Despite the importance of resolving reaction kinetics for understanding the processes of prompt CWA defeat,only limited related information could be found in the literature,where most work focused on processes important for slow decontamination.

Relevant results on initial removal by adsorption of DMMP on aerosolized TiO2have been described in Refs.[19,49,86].In experiments,ambient humidity was varied and UV activation of the TiO2surface was explored.Time-resolved gas phase concentration of DMMP in the reactor chamber shown in Fig.6a is presented in Fig.15.In the experiment shown,the humidity was 4%,and no UV radiation was used.Aerosolized TiO2(anatase)particles were injected into the chamber filled with DMMP vapor.The aerosol was introduced for a period of 10 min at a known rate while the gas phase concentrations of DMMP and methanol were monitored.Within the first minute of the aerosol injection,the concentration of DMMP dropped by 96%.After a period of rapid decline,concentration decreased steadily over several minutes.Methanol was detected as the only reaction product.The concentration of methanol was highest during the first minute;it then decreased gradually.

Fig.15.Vapor phase concentration of DMMP and methanol in a chamber at 23°C,4%relative humidity during injection of titanium oxide aerosol.The solid line for DMMP shows the function(a double exponential)used to describe the data in Ref.[86].

This measurement allows one to evaluate initial removal rates of DMMP.The time resolution and the accuracy of the concentration measurements are not sufficient to resolve individual reaction steps,i.e.,to separate adsorption from any subsequent reaction of the adsorbed species.Nevertheless,the bulk initial removal rate of DMMP from the gas phases can be useful for understanding the reactions assisted by the metal oxide powder.Taking the removal process as first order in DMMP concentration and in catalyst amount as the simplest assumption,the change in the gas phase concentration of DMMP,CDMMP,can be described by:

wherekis the effective rate constant,moxis mass of oxide andtis time.The value ofkcan be found using the functions forCDMMP(t)andmox(t)reported in Ref.[86].This value,in units of min-1·g-1(per g of metal oxide)is shown as the dotted line in Fig.15.The effective rate diverges initially due to the functions chosen in Refs.[86]to representCDMMP(t)andmox(t)based on the measurements,but within the first minute of reaction,it stabilizes neark=2.5 min-1g-1,or 1.3·10-4s-1m-2before dropping off.Values are given here per unit mass,or unit surface area of the TiO2particles,using the reported speci fic surface area of 320 m2/g.The effective rate constant obtained for the short plateau characterizes the bulk decomposition rate of DMMP by the used titania powder.The following drop of the effective rate constant is observed after the majority of the DMMP has been removed from the gas phase.Slowing of the DMMP removal due to saturation of the TiO2surface can be ruled out,because TiO2particles continue to be injected in the chamber at a similar rate up until 10 min.One can speculate that mass transfer processes become more prominent as the overall gas phase concentration of DMMP decreases leading to its smaller concentration gradients near the TiO2particle surfaces.

Decomposition rates at higher temperatures can be determined to some extent from the packed bed/flow reactor experiments described in section 2.2/Fig.6b if reactor species balances and catalyst deactivation laws are combined.As an example,in Ref.[53],DMMP vapor was passed over a Pt/Al2O3catalyst,which was found to gradually lose activity.The time-dependent DMMP conversion observed at different temperatures was combined with the empirical catalyst deactivation law to extrapolate back in time to the state of the fresh catalyst surface.This initial catalytic DMMP decomposition rate was found as:

Unfortunately,this rate is not correlated with any speci fic catalyst characteristic,such as mass,surface area,or site density.Therefore,it cannot be used for comparisons between catalysts used in other experiments.The authors compared the Pt/Al2O3catalyst used with the same reactor filled with glass beads assumed to be catalytically inactive.The reported thermal DMMP decomposition rate was found to be lower:

The activation energy for thermal decomposition identi fied in Eq.(4)is only marginally higher than for the reaction with catalyst(see Eq.(3)),leading the authors to suggest that there is some catalytic reaction occurring even in their inert case.In addition,that study used porousγ-alumina as catalyst support,and examination of results for different catalyst particle sizes led the authors to conclude that the overall process is likely limited by pore diffusion rather than any process occurring at the catalyst surface.No quantitative comparison between thermal and catalytic decomposition,and between different catalysts is therefore feasible based on these results.Therefore,quantitative measurements for the purpose of prompt removal or inactivation of DMMP by catalytic activity in combustion clouds should focus on nonporous catalysts.

3.5.DFT studies:reaction coordinate

Computational studies have primarily focused on modeling decomposition of DMMP and sarin at room temperature,where the reaction can be reduced to cleavage of an alkoxy group or fluorine attached to the phosphorus center.For example,the DFT-predicted enthalpies of adsorption/desorption and activation energy barriers for decomposition of sarin and DMMP over a pristine and hydroxylated zinc oxide surfaces are shown in Fig.16.The similarity in the predicted energy states for adsorbed and decomposed sarin and DMMP molecules is apparent.

Calculations show that for both sarin and DMMP,adsorption to hydroxylated surface is a less exothermic process compared to pristine surface of ZnO.The strength in adsorption to an acid site compared to hydroxyl is well understood[61],however a unique prediction made by the model is that hydroxyl sites are capable of adsorbing as well as decomposing attached molecule[116].The effect of hydroxyl groups in reducing the activation energy for decomposing DMMP was also reported in other DFT studies[80,117].Further decomposition over hydroxylated surface,yielding surface-bound IMPA and fluorine for sarin and MMP and methoxy group for DMMP,occurs with a lower activation barrier than respective decomposition reactions over pristine surfaces.

Not presented in Fig.16,desorption of surface bound species(e.g.,IMPA/F and MMP/methoxy)is predicted to be endothermic,suggesting that once formed,these species will remain bound to the surface at room temperature.

Fig.16.Energy states of 1st step decomposition of(a)sarin and(b)DMMP over(1010)zinc oxide surface calculated using PBE functional[116,117].

A summary of DFT-calculated adsorption enthalpies and energy barriers for decomposition of sarin and DMMP over some oxides is given in Table 3.All energies are expressed with respect to the adsorbed state.The predicted reaction mechanisms are generally similar to those observed in experimental spectroscopic studies.For example,the P-F bond in sarin was observed experimentally to be cleaved upon interaction with a magnesia surface[32].This is also predicted computationally[109].Copper oxide cleaves methyl and methoxy groups from DMMP as predicted and observed experimentally in Refs.[65].Similar to the thermal decomposition[12,53,126],energy barriers for decomposition of sarin are lower than for DMMP on the same oxide surfaces.In both experiments and calculations,reduced oxide surfaces,such as that of molybdenum oxide,are more reactive than pristine surfaces.While most predicted reaction trends are in line with experimental observations,there is not enough experimental data to validate the predicted activation energies and enthalpies of adsorption and desorption for either sarin or DMMP.

Table 3 Energy states for decomposition of sarin and DMMP over various surfaces calculated using DFT.

4.Discussion

To assess the viability of decomposing CWA using metal oxide aerosol generated by combustion in a prompt defeat scenario,a simple estimate can be made for the minimum concentration of metal oxide required to decompose CWA faster than it is decomposed thermally.Unfortunately,no reliable data describing heterogeneous kinetics of CWA or CWAS decomposition could be found in the literature.It is,however,possible to make an estimate assuming that the mass transport of CWA in the gas phase to the surface of oxide particles is the rate-limiting process.Note that mixing of the aerosol and CWA is not considered.Consider an environment containing homogeneously mixed aerosol and CWA with a concentrationXCWA[mol/m3].For example,the value ofXCWAcan represent the toxic concentration of sarin shown in Fig.1,CWA will be destroyed thermally at a rate described by Glaude et al.,[126].Respective thermal decomposition kinetics identi fies the characteristic reaction time as a function of temperature,tth(T).This time can be defined as required to reduceXCWAbynorders of magnitude.In a prompt defeat scenario,at the same time and temperature,the CWA can be destroyed by reacting with an aerosolized metal oxide that was generated at a concentration ofY[g/m3]by combustion of metal fuel added to an explosive charge.The metal-oxide assisted decomposition of CWA should occur faster than its purely thermal decomposition in order for the oxide to have an effect.Thus the characteristic time of the oxide-assisted decomposition,toxshould be:

In the case of gas phase diffusion serving as the rate-limiting process,the characteristic timetoxcan be taken as the time required for a CWA molecule to diffuse over the distance separating aerosolized oxide particles,l:

whereDis the diffusion coef ficient of CWA in air.Taking into account Eq.(5),we obtain

Assuming constant pressure,the diffusion coef ficient is expressed using the Chapman-Enskog theory as:

whereDis in(cm2/s),indices 1 and 2 are for diffusing molecule,CWA and bath gas,air,respectively,Mis the molar mass(g/mol),pis the pressure(bar),σ1,2is the average collision diameter(?)andΩ is a temperature-dependent(dimensionless)collision integral.The diffusion coef ficient is calculated for various temperatures using the approach described in Ref.[127].For calculations,properties for CWA were taken for sarin from Ref.[128,129].Data used and additional details are given in the Supplement.

From characteristic distance separating aerosol particles,it is easy to estimate the volume surrounding each oxide particle:

Thus the minimum mass concentration of nonporous oxide particles with mass,m,required to decompose CWA faster than it is decomposed thermally will be:

whereρa(bǔ)ndrare respectively density and radius of the oxide particle.From Eq.(10),it is clear that generating finer aerosolized smoke particles leads to smaller required mass concentrations of the aerosol,for which the rate of gas diffusion-limited oxideassisted decomposition of CWA matches the rate of its thermal decomposition.Taking the metal oxide particles forming in combustion products to vary in size in a characteristic range of 0.1-1μm[130,131],respective values forYmincan be calculated for each particle size as a function of temperature.For each calculation,the time,tthcan be taken for different extent of thermal decomposition of CWA,e.g.,for 99%and 99.9%destruction of the initial CWA concentration(e.g.,n=2 or 3)for the practical range of temperatures,e.g.,500-1400°C.The characteristic times and respective values ofYminare shown in Fig.17 for particles with radii of both 0.1 and 1μm.The range of temperatures considered here is based on the respective predicted thermal CWA decomposition times,roughly varied from 1μs to 100 ms.In its turn,this range of times is associated with prompt agent defeat applications,targeting CWA escaping the high-temperature region of a fireball.

If concentrations of the aerosolized particles formed in combustion exceedYminvalues shown in Fig.17,the rate of metal-oxide assisted decomposition of CWA may exceed the rate of its thermal decomposition for each temperature.The values ofYminshown in Fig.17 account only for rates of gas phase diffusion(as a function of temperature and properties of the diffusing molecules)and thermal decomposition of CWA;these values are completely independent on the type of metal oxide.In other words,the metal oxideparticles are assumed to decompose CWA molecules instantly upon contact and with inde finite decomposition capacity.While the rates of heterogeneous oxide/CWA reaction are poorly defined based on the previous work,it has been shown that the capacity of metal oxides is limited.The BTP as determined in flow reactor experiments(see sections 2.2 and 3.2)can be taken as a measure of the catalyst capacity and can,in the first approximation,serve as a useful reference for assessing potential of different oxide particles for prompt defeat applications.In other words,BTP associated with the chemical properties of the oxide is the only material-speci fic characteristic considered for this preliminary analysis.If the concentration of CWA is,the mass concentration of aerosol required to decompose it rapidly can be obtained as:

In Eq.(11),the value ofBTPmis taken per unit mass of the oxide,as reported in the literature[92,96].Applying Eq.(11)assumes that the aerosolized smoke particles have the same speci fic surface as the oxide used in the measurements(typically involving a tubular flow reactor,Fig.6b),in which BTP was determined.However,as discussed above,often differently prepared oxides had different surface areas available for heterogeneous reaction with CWA or CWAS(DMMP).The values of BTPAnormalized per such surface area were calculated and shown in Figs.12 and 13(for DMMP).Assuming that those values remain characteristic of speci fic oxides,an estimate different from that given by Eq.(11)can be obtained.The mass concentration required to decompose the CWA by aerosolized powders with different particle sizes can be calculated instead as:

whereAs=is speci fic area of the powder,affected by its particle size.Forγ-Al2O3particles with radii of 0.1 and 1μm considered here,the values ofAsare 8.2?104and 8.2?103[cm2/g],respectively.

For bothYmandYA,the concentrations may somewhat decrease with temperature based on the observed increase in BTP with temperature(Fig.12);however this effect is relatively weak and poorly known for the range of temperatures shown in Fig.17;therefore,it is presently neglected.

For comparison with the minimum calculated concentrations,Ymin,the value ofYmand two values ofYAfor particles with radii of 0.1 and 1μm are shown in Fig.17.For this estimate,it was assumed that for sarin,,consistent with the data shown in Fig.1.The values of BTPmand BTPAare taken forγ-Al2O3(cf.Fig.13).A separate estimate forYmis also shown where BTPmis taken for a mixed Al2O3/CuO oxide.Recall that the BTP values used here were reported for DMMP;thus,the present analysis would need to be corrected using respective BTP values for sarin,which are not as well defined as those for DMMP.With the above caveat,the results in Fig.17 show that when the concentration of the aerosolized smoke particles exceeds bothYminandYm(orYA)the aerosolized smoke particles are expected to be effective in terms of both the rate of CWA decomposition exceeding that of its thermal decomposition,and the capacity required to destroy the toxic concentration of CWA.

Fig.17.Characteristic times of thermal decomposition of sarin following Ref.[126](a);and estimated mass concentrations of aerosolized smoke particles required to decompose sarin at different temperatures(b).Unless labeled otherwise,all values of mass concentration are estimated forγ-Al2O3.See text for details.

For example,relying on the value ofYAand considering smoke as aerosol ofγ-Al2O3particles with radius of 0.1μm,one obtains that the smoke mass concentration should exceed 1.56?10-8[g/cm3]to destroyof sarin.The horizontal line representingYAfor this case crosses the respective line ofYmin(e.g.,fortth,2)at about 778°C.Thus,if the actual aerosol concentration is determined from BTPAand isYA,the aerosol will remain effective in destroying sarin faster than it is decomposed thermally only up to 778°C.Similarly,based on BTPA,for 1-μm radiusγ-Al2O3particles,the minimum smoke mass concentration should be an order of magnitude higher,1.56?10-7[g/cm3],to destroy the same concentration of sarin.In this case,the horizontal line representingYAcrosses the respective line ofYmin(also fortth,2)at even lower temperature of 626°C.This determines an even lower maximum temperature for which the 1-μm aerosolized alumina particles with the concentration required by BTPA,remain effective in destroying sarin faster than it is decomposed thermally.Clearly,higher concentrations and smaller size of aerosolized smoke particles will lead to faster reactions,greater cut-off temperature,and greater total capacity of the aerosol for decomposing CWA.

The line forYmshown in Fig.17 for the mixed Al2O3/CuO oxide,for which BTP is an order of magnitude greater than forγ-Al2O3suggests that an order of magnitude lower smoke concentrations are required for CWA defeat if that smoke comprises mixed oxides.Similar estimates for values ofYAshift them down from the respective horizontal lines forγ-Al2O3,just like forYm;these additional lines are omitted in the already busy Fig.17.Note that while an increased BTP shifts the required oxide concentration down,it also results in the respective reduction in the maximum temperature,at which the oxide-assisted decomposition of CWA remains faster than its thermal decomposition.

Finally,a concentration of aluminum oxide smoke[132]is shown in Fig.17 estimated based on the data reported in a recent experimental study using a 243 L chamber for experiments on prompt defeat of biological weapon agents[133];similar equipment can also be used for CWA agent defeat studies.In those experiments,the charge of aluminum powder of 7.5 g was used,which could generate up to 14.2 g of alumina[132].This leads to a value ofYexp=5.83·10-5.As is clear from Fig.17,this smoke concentration is well aboveYAestimated as required for CWA defeat based on particle sizes as large as 1μm.

The analysis presented in Fig.17 can be reduced to a preliminary practical recommendation shown schematically in Fig.18.Intersections of horizontal lines for various values ofYAwith respectiveYmincalculated for the same particle size are shown as minimum required smoke particle concentrations as a function of the particle size.These points are calculated comparing equations of(10)and(12)and solving for the particle radius:

Fig.18.Minimum concentration of the aerosolized smoke particles of different sizes required for them to be able to decompose CWA faster than it is decomposed thermally up to the temperature shown.

The shaded part of the plot shows the range of concentrations that can be effective in prompt destruction of CWA depending on the aerosolized smoke particle size.The range of particle sizes considered:0.01-10μm covers all possible smoke particle sizes expected in experiments.The second line in Fig.18 shows the maximum temperature,at which particles of given sizes introduced at the minimum concentration remain more effective in decomposing CWA than purely thermal decomposition.Of course,concentrations higher than the minimum would lead to greater maximum temperatures.The plot shown in Fig.18 was prepared using BTPAreported for decomposition of DMMP onγ-Al2O3,diffusion rates for sarin in air,andtthrepresenting 99.9%of thermal decomposition of sarin.Changes in BTP or diffusion rates will change speci fic size-dependent concentrations required for different temperatures;however,conceptually,such plots can be used for designing energetic systems generating appropriate amounts of chemicidal smoke.

Reducing the smoke particle size appears to be an effective strategy of increasing its chemicidal ef ficiency.Conditions leading to formation of fine oxide smoke and fibrous surfaces during combustion have been identi fied for metals including aluminum and magnesium(typical fuels)and molybdenum,iron,and tungsten(typically used as respective oxides in thermite compositions)[134-137].Modifying the energetic composition with the goal of approaching respective conditions may thus be desired to produce such fine particles.

The results discussed here show how a greater catalyst capacity,experimentally determined as BTP,leads to a greater ef ficiency of oxide-assisted decomposition of CWA,and thus to reduced smoke concentration required for prompt CWA defeat.Mechanistically,increasing BTP for a metal oxide involves both accelerated adsorption of CWA molecules to the oxide surface and their rapid destruction.Studies focusing on adsorption of DMMP to metal oxide surfaces(illustrated above in Figs.7 and 9 as well as DFT studies illustrated in Fig.16)suggest that chemisorption to Lewis acid sites becomes more prominent at greater temperatures and leads to a more effective further decomposition.Respectively,it is desired to produce oxides with active Lewis sites as combustion products;γ-Al2O3can be treated as a useful starting material with such properties.Oxides with greater acidities,such as MoO3may be of interest,although formation of such oxides in combustion is not energetically attractive.

Spectroscopic studies on copper oxides indicate potentially higher BTP adsorption capacities than for non-reducible oxides[65,78].An increased BTP can be achieved,as shown in Fig.13,by producing ternary metal oxides.Indeed,an approach in which alumina,a common combustion product of aluminized propellants,is coupled with oxides of copper,molybdenum,etc.,can be considered.One method attractive for generation of such ternary oxides and mentioned above involves using thermite compositions as explosive additives.Metal oxide oxidizers used in such compositions commonly involve CuO and MoO3[138-141].Upon oxidizing aluminum,reduced metals can re-oxidize when they are exposed to a gaseous oxidizer[141].Although this secondary oxidation will not contribute signi ficantly to the heat release of the explosive charge,it may lead to the formation of the desired ternary oxides with greater BTP leading to a more effective oxide-assisted decomposition of CWA.

A more rapid CWA decomposition pathway involving reducible oxides,capable of donating oxygen,as previously identi fied and illustrated in Fig.10 provides further support for employing thermite compositions,containing readily reducible metals,which can oxidize when exposed to the oxygenated environment.

This literature review identi fies the lack of data and measurements quantifying kinetics of CWA decomposition on metal surfaces for the range of temperatures and times appropriate for prompt defeat scenarios.Such mechanisms must be addressed in future research.In particular,experiments building on studies as in Ref.[86]Mre desired,involving decomposition of CWAS on aerosolized metal oxide powders.Both exposure time and temperature must be varied systematically and the products should be analyzed,including the decaying concentration of CWAS as well as generated new species.

When studying oxide-assisted reactions involving decomposition of CWA,formation of oxidized products,such as CO and CO2(see Fig.14)should be accounted for.Speci fically,exothermic reactions can lead to a local temperature increase[12],heating up the oxide particles aerosolized in air,and thus leading to changes in the further reaction kinetics and,possibly,in BTP.Such temperature increase may need to be considered for oxide particles that have substantially cooled after exiting the fireball.Note that the concept of BTP can only be used for approximate estimates,as done above.It cannot apply to calculations accounting properly for the kinetics of CWA decomposition on the metal oxide surface,which is expected to become available and guide further development of reactive materials for prompt CWA defeat munitions.

5.Conclusions

Decomposition of gaseous CWA on surfaces of metal oxides has been studied mostly to identify the reaction characteristics at low temperatures and extended exposure times,relevant for decontamination of affected structures.Most experimental efforts relied on measurements from different types of tubular flow reactors.Adsorption and reaction mechanisms were studied primarily relying on spectroscopic measurements involving decomposition of CWA and CWAS on surface of metal oxides packed as porous solid.DMMP has been the most common CWAS for sarin;multiple studies involving DMMP decomposition on surface of diverse metal oxide based materials have been reported.The mechanisms of DMMP adsorption and its initial decomposition steps on different oxides were characterized and found to be qualitatively similar to such steps for sarin.DMMP molecules need to be chemisorbed to a Lewis acid site prior to being decomposed on the oxide surface.A reaction sequence involving formation of MMP and MP occurs for many non-reducible metal oxides.For reducible oxides readily generating lattice oxygen,the decomposition accelerates at low temperatures involving several parallel reactions.These initial reaction steps observed experimentally have been also described theoretically using DFT.Calculations determine activation energies required for different reaction steps.These can be used in the future to predict the respective kinetics.Further reaction steps in oxidizing environments involve exothermic formation of oxidized carbon-containing species.The associated heat release may raise temperature of the metal oxide particles participating in the reactions;this needs to be accounted when developing respective prompt CWA destruction models.Unfortunately,very limited information is available for the rates of heterogeneous CWAS decomposition on surface of metal oxides.Future experiments are needed to quantify such rates for the range of characteristic times and temperatures of interest to prompt CWA defeat scenarios.Existing measurements quantifying breakthrough point(BTP)for many metal oxides are available;however,the data are limited to relatively low temperatures and describe oxides based on their mass.Thus,the effect of the available oxide surface area remains poorly understood;in many present experiments,the CWAS diffusion in the oxide pores could have been a deciding factor for BTP.In a prompt defeat condition,the oxide is not expected to be porous.It is expected to be present as an aerosolized smoke;thus characterizing the particle size distribution and respective speci fic surface area of such a smoke is important for assessing its relevant BTP.The value of BTP can serve to estimate preliminarily the required mass concentration of the aerosolized smoke particles necessary to decompose CWA present or released in air at a speci fic concentration.Additionally,the metal oxide assisted decomposition of CWA can be more effective than its thermal decomposition within a limited range of temperatures.For a given temperature,the concentration of smoke particles available should be sufficiently high to at least ensure that the diffusion transport of CWA molecules to the surface of particles occurs faster than purely thermal decomposition of these CWA molecules at the same temperature.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to in fluence the work reported in this paper.

Acknowledgment

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

Appendix A.Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.dt.2020.08.010.