Annular Reactor

8/8/2019 Annular Reactor

1/10

1

P

Photocatalysis for the Destruction of Aqueous TNT, RDX, and HMX

Steven K. Showalter, Michael R. Prairie, Bertha M. StangeSolar m erm al Technology Department 6216

Sandia National Labora tories, Albuquerque, iU.4 87I85-0703

Philip J. Rodacy and Pamela K. LeslieExplosive Subsystems Department2652

Sandia National Laboratories, Albuquerque, N M 87185-0329

AbstractThe photo-destructiono , igh explosivesHMX, DX and TNT w as investigated using two

systems (Ozone Versus titanium dioxide),two reactors b o t vs annular reactor), andtw o types of lamps(Iooo w a n H g - xe VS 25 wat t Lp Hg). A mass balance was performed on reactions executed underPseudo-solm cond itions, and relative reaction rates and products were com pared for Ozone and titaniumdioxide based Processes. The ratiosOf relative product formation is also discussed. Results show thatthere Was little diffh-ence n the reactions performed in the annular reactor when either Ozone or titaniumOxide Were used. The chemistry0f-X and HMX re very similar, as expected. Fu ture w ork involvingthe mechanism isalso discussed.

DISCLAIMER

This report was prepared as an account of work sponsored by an agency of the United Stat esGovernmen t. Neither th e United State s Government nor any agency thereof, nor anyof theiremployees, makes any warranty, expressor implied, or assumes any legal liabilityor responsi-bility for the accuracy, completeness,or usefulness of any infomation, apparatus, product,Orprocess disclosed,or represents that its use would not infringe privately owned rights. Refer-ence herein to any specific commercial product, process, or service by trade name, trademark,manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recom-mendation, or favoring by th e United States Governmentor any agency thereof. The viewsand opinions of authors expressed herein do not necessarily state or reflect those of the

United State s Governmentor any agency thereof. ~~~y~~~~~~~~QF THIS OaeiiMEkT 'S U N L f M~ ~~-


2/10

DISCLAIMER

Portions ofthis

document may be illegiblein electronic image products. Images areproduced from the best available originaldocument.


3/10

IntroductionW e have been investigating TiO,

photocatalysis for a numberof years forapplications involving waste waterremediation.3 Recently, our attention has

become focused on the problems of aqueousmunitions ~ a s t e . ~ ' ~

explosives originates from numerous munitionsprocesses including load and pack operations,explosives machining, cutting and hogging outopera tions, as well as demilitarization activities.Water containing 2,4,6-trinitrotoluene (TNT)that has been exposed to light is commonlyreferred to a s 'lpinkwater". This 'lpinkwater'lcontains not onlyTNT (and its degradation

Aqueous waste containing dissolved high

RDX

O2

N02

TNT HMX

Figure 1 Structure ofH E Test Materials.

byproducts), but may also contain otherexplosives and contaminants as well (High meltexplosive, HMX, or research developmentexplosive,RDX). The actual makeup of thewaste wa ter is dependent on the type of processthe effluent is from. These dissolved materialsmake the water highly toxic and thereforeuntrea table via classic biological waterpurification systems. Hence treatment isrequired before the m aterials can be dischargedto the environment or municipal sewage streams.Traditionally the water is treated by filtration toremove suspended solids, then passed overactivated carbon t o remove the organics. This,however, has some disadvantages. The practiceof using the carbon only once can be expensiveand contribute to air pollution since the spentmaterial is openly b ~ r n e d . ~arbon regeneration

is feasible in som e cases, although ultimately thecarbon is again disposed o f through openburning. Changing environmental regula tionsindicate that open burning will be prohibitedwithin a number of years. Furthermore, newly

proposed environmental discharge criteria willnot be met through carbon adsorption systems.Thus, new, low-cost technologies thatdo notgenerate secondary waste are being sought fo rthe treatment of this "pinkwater" waste.

We are currently investigating twotechnologies to achieve this goal: titaniumdioxide (TiO,) photocatalysis and photolyticdegradation through the action of ultravioletlight OTV) and ozone ( 0 3 ) . hotocatalysis withTiO, semiconductor occurs when light causes

the formation of electrodho le (e-/h+)pairs inparticles suspended in solution. The e'/h' pairscan react independently at the surface of thecatalyst particle with solvent or contaminants, or

Oxidation

.-0 2 0 2

lN T Ueducthproduct.

Reduction

Figure 2 Oxidation and Reduction on TiO,.

can recombine internally to generate heat. Theholes react with water to produce hydroxylradicals (OH') which are capable of oxidizingorganic contaminants to carbon dioxide, waterand dilute mineral acids. The elec trons can reactwith either dissolved oxygen to generate fbrtherOH' or can directly reduce some species insolution such as m etals o r organics (ie.nitrogroups to amines).


4/10

Titanium dioxide is the most commonlyused photocatalyst due to its stability, lowtoxicity, low cost, and effectiveness under nearUV light (< 390 nm). Since nearUV is presentin normal sunlight, TiO, can be used without the

requirement of electrically powered lamps.Semiconductor catalysts have an advantage inthat both oxidation and reduction can beperformed on the surface of the same material.

Ozone can be typically produced using185 nm light in the presence of oxygen.Decomposition of the0, molecule can beinitiated by either a collision with a surface orabsorption of254 nm light. These reactionsproduce oxygen radicals, which in the presenceof water will generate hydroxyl radicals. Ozone

can also react d irectly with organiccontaminants, how ever due to its normally lowconcentration and short lifetime, this is not theprimary pathway o f ozone decomposition.Draw backs of ozone treatment are the need forartificial lamps, no reduction chemistry, and thetoxicity of the active molecule.

ExperimentalSamples of synthetic and actua l

explosives-contaminated water were used in aseries of treatability tests. Laboratory syntheticwastewater was prepared from deionized waterand the appropria te high explosive. Theconcentrations of the high explosivesapproached the ir saturation limitsin water: TNT(90 m a ) , RDX (40 mg/L) and HMX (4 m a ) .All of o ur simulated solutions are colorless tobegin with and stored in amber glass away fromlight. Actual pinkwater was obtained from theLouisiana Army Ammunition Plant(LAAP),Thiokol Corporation, S hreveport,LA . Thiscontained 63 mg/L of TNT, 5 1.5 mg/L RDXand a number ofUV created photointermediates.The color in this case is deep pink ororange.

used in these reac tions. The first involved400Three types of experimental setups were

mL batch reactions in a glass pot reactor(capacity 1.5L) illuminated with a 1000 WattHg-Xe ozone free arc lamp (see Figure3) . Theresulting spectrum of lightfrom this lamp is verysimilar to sunlight.6 All optics employed w ere

constructed of quartz in order t o allow low-losspassage of both visible andW light. The potwas sealed with a quartz lid and a Viton O -ring.Anaerobic reactions were carried o ut bysparging the reactor with nitrogen gas though afritted glass tube. All reactions were stirredusing a magnetic stirrer and samples werewithdrawn from a sea lable sampling port.Titanium dioxide was used in this se tup tosimulate sunlight acting on a photocatalyst.

- oPmver

waterollt

+

Water

Figure 3 1000 Watt Pot R eactor.

The second apparatus was an annularcore recirculation reactor with1 L of solutionpassing in front of a low pressure25 Watt quartzjacketed Hg a rc lamp (see Figure4).7 This lampwas constructed of fbsed quartz and emittedprimairily at 254 nm (ca. 95-98%), with a smallamout of emission at 185 nm. The spacebetween the walls of the reac tor and lamp wereblanketed with argon. The interior wall of the


5/10

LAmonorAir J 4

Figure 4 NREL Continuous Flow R eactor.

reactor was com posed of k s e d qu artz and thebalance of the system was built of Pyrex andteflon tubing. This reactor was used withtitanium dioxide.

the annular reactor but w ith a change in thesetup so that a stream of air at60 mL/minpassed betw een the wall of the reactor and thelamp. This gas stream was then bubbled into thesolution.No catalyst wa s used in th is lastreactor, since here our desire was t o simulateozone photodegradation in the absence ofcatalyst. Reactions were run for5 hrs andcooled to 20" C witha water jacket. Reactionsinvolving TiO, were filtered before analysisusing a 10mL disposable syringe and a0.25 pmTeflon syringe filter.

Tioxide Tilcom HACS (210 m2/g) powdersuspended at 1 g L n the 1000 Watt case and0.25 g L n the 25 W att case. The lower loadingwas used in the annular reactor based on resultsfrom the National Renewable Energy L aboratory(NREL) where th e reactor was designed andbuilt.'

using high performance liquid chromatography(Waters HPLC with 231 nm W detector), ionchromatography (Dionex IC with conductivitydetector), and tota l organic carbon analysis(Shimadzu TOC). UV Nisible spectroscopy wasused t o provide qualitative information on

The third reaction was also performed in

The photocatalyst was anatase TiO,

All reactions were sampled and analyzed

reaction progress in the caseof the "pinkwater"reactions.

Results and DiscussionCarbon and n itrogen mass balances were

performed independently on all three explosivesin the 1000 Watt reactor under aerob ic andanaerobic conditions (see Tables 1,2,3).

In the case of TNT, the d estruction ofthe explosive was less com plete under nitrogenthan in air (Table 1). This is to be expectedifthe initial reactionis oxidative in nature.However, a smaller amo unt of solubleintermediates, as well as a low er overall numberof individual species(5 vs . 3) are formed underreducing conditions. This may have the addedbenefit of reducing the toxicity o f the solution(Table 4).

At the end of the anaerobic reaction thecatalyst is coated with a brown su bstance(60%deposited carbon,45% nitrogen) or a3/1 bymass ratio of carbon to nitrogen. This resultwhen compared to a 2/1 ratioof carbon tonitrogen in the parent com pound shows that thisis definitely not TNT, but rathe r som edecomposition product. Analytical techniqueshave yet to positively determine the nature ofthis coating, but early results point towards apoly-aniline analogue produced through thereductionof nitrogroups to amines or diazolinkage^.^$'We have also found that the surfaceof this coating, as observed with X-rayphotoelectron spectroscopy(XPS), differs fromthe bulk, indicating that there isa change in thedeposited species over the course of thereaction. 1

also has some detectable amountsof carbon andnitrogen (15% and 2% respectively, with a ratioof 15/1 by mass), however no color is evident onthe surface. Here it wou ld seem thatdecomposition of the nitrogroups w ould be animportant stepin the destruction of TNT underoxidative photocatalytic conditions .

The catalyst from the aerobic reaction


6/10

Significant mineralization of the highexplosiveis realized under these conditions bylooking at the amo unt of unaccounted for carbon(45% of total carbon).

~~

ProductsUnaccounted CH E C

Table 1 TNT Mas s Balance Po t Reaction

TNT.N, TNT.Air

% %

4.48 47.57

0.34 0.00

lesults

~

Inorganic CC Intermediates

0.45 0.0032.93 37.76

C on TiO, I 61.81 I 14.67 I

69.74

0.002.03

12.36

0.16

15.71

Experiments using cyclingof first reductive and

then oxidative conditions in TNT w astewaterhave shown this to be an effective way to cleanoff the catalyst, while producing fewerintermediate species. This does have thedrawback of increasing the reaction time, butmay prove u se fil in decreasing the overalltoxicity o f the treated water, if biotreatment isdetermined to be a useable end treatment for thisw a ~ t e . ~ ~ ~ ~urther experiments will determinequantitatively the degree o f toxicity/mutagenicityfor the two m ethods.

The reactions involving RD X also had ahigher conversion percentage under air thanunder nitrogen (Table2). How ever, the overallconversionis disappointing under the bestconditions(35% with N,, 70% with air). Theamount of material deposited on the catalyst isindependent of the presence of air and is entirelycarbon based. The number of intermediates

formed under aerobic conditions is4 times thatcreated under anaerobic conditions(2 vs 8),although this could bea fbnction of the muchlower conversion under nitrogen andcorresponding lower side product reactions.

the same pot reactor have shown that the use ofanaerobic conditions accelerate the degradationof RDX (vs RD X alone) while still completelydecomposingTNT in 300 minutes. Thisisassumed to be related to th e presence o f havinga co-reactant in solution to electrochemicallybalance the e- h+pairs in titanium dioxide. Therate of the TNT decompositionis unaffected bythis addition. Only five detectable intermediates

Reactions involving TNT andRDX in

Table 2 RDX Mass Balance Pot ReactionResults

I IRDX. N, IRDX.AirProducts YO %Unaccounted C 0.00 41.76

are observed in the final sample, vs seven

intermediates for the two reactionsindependently (Note: there was no observableoverlap of intermediates).

The HMX reactions indicate that aerobicconditions are the best in effecting mineralizationwith 40% conversion underN2 and 100% underair (Table 3). Very few interm ediates areformed that are detectable, but this is


7/10

understandable given that theHM X reactionsare based on only4 mg/L of explosive and thatthe intermediates may no t be present in highenough concentrations to be detectable. Thematerial deposited on the catalyst is almost

exclusively carbon in both cases. This isinagreement with theRDX data. We expect thatthe chemistry between these tw o will be verysimilar under th e sam e conditions.

Table 3 HMX Mass Balance Pot ReactionResults

HMX.N, =.AirProducts % %Unaccounted C 31.42 80.75

61.92C on TiO,

Unaccounted N 29.88 57.14H E N 6 1 . 9 2 0.00N on TiO, I 0.28 I 0.00 IN asN 0,- 19.54N as NO;N a s N H ' 2 3 . 3 3

100

80

.5 600)c.-

E[IL 40s

20

00 50 100 150 200 250 300

Reaction Time (min)

1- TNT TOC RDX TOC HMX TOCl

Figure 5 Normalized TOC forHE with TiO,.

0 10 20 30 40 50 60Reaction Time (min)

IB TN T HPLC 8 RDX HPLC A HMX HPLCl

Figure 6 NormalizedHPLC or HE with TiO,.

It would be desirable, based on the aboveresults, to employ cycling of the reactionbetween oxidative and reductive conditions toobtain the most effective destruction of mixedwaste involving the three listed explosives. Thisis based on the acceleration of destructionofRDX with TNT under anaerobic conditions,while the catalyst is maintained in a m ore usefulform when it is used under oxidative conditions.In order to determine that the results were notreactorAamp specific andto have the freedom ofusing ozone as an oxidant, we also used asecond reactor system. This modified 25 WattNational Renewable Energy L aboratory(NREL)annular core reactor was continuouslyrecirculated mechanically, rather than beingstirred with a m agnetic stirrer. The reactions inthis system were performed for comparisonpurposes of the reactivity of titanium dioxide vs.ozone. The reaction results were all very similarwhen comparing ozone generation in-situ againstitanium dioxide photocatalysis with this system(Figures 5-8). In the reactions involving TNT,there is a slight differencein the amount ofcarbon in solution at the end o f the reaction(85% drop in TiO, vs 78% drop in ozone,Figures 6 and 8). This is attributed toadsorption of the carbon on to the surface of the


8/10

catalyst. The o zon e reaction is slightly faster, butfour intermediates are present at the end of thereaction, whileonly two are present whentitanium dioxide was used (Table4).

Table 4 Conversion and Intermediates ofHEReactions in IConditions

TNT/N,/TiO,

TNT/Air/TiO,

TNT/Air/TiO,TNT/Air/O,

RDX/N,/TiO,

RDX/Air/TiO,RDWAir/TiO,

RDX/Air/O,HMX/N,/TiO,

HMX/Air/TiO,

HMX/Air/TiO,HMX/Air /O,

fferent AppaiLampReactor

Hg-Xeff ot

Hg-Xeff ot

Hg/Annular

Hg/Annular

Hg-Xe/Pot

Hg-Xe/Pot

Hg/Annular

HglAnnular

Hg-Xeff ot

Hg-Xe/Pot

Hg/Annular

Hg/Annular

tus

% HE Conv.

99.3

100

100

100

34.8

69.7

100

100

38.1

100

100

100

# of Intel

5

3

2 4

2 8

1 1

3

2 1 1

-

-In both cases, the clear solution turnedimmediately red following ignition of the lampand then slowly changed to lighter orange andyellow colors over the course of the reaction.

The RDX and HMX eaction rates,

100

80

.$ 60

IY 40

..................................................

cnS.-

...............................

Es


I* NT TOC e DXTOC A HMXTOCI

Figure 7 Normalized TOC forHE with Ozone.

intermediates, and final total carbon w ereinsensitive to the type of oxidation method used.This demonstrates, as shown in the Xe-Hglamppot reactions, that RD X destruc tion is relativelyinsensitive to reductive chemistry. Based on the

similarity of the moleculeHMX with RDX , thechemistries should be analogous and we find tha tindeed this is the case. One o f the problems ofworking withHMX is that most o f our analysistechniques become limited af ter the first fewminutes of the reaction. Since RD X andHMXseem to behave similarly,in kt ur e work we planto use RDX and extend the work toHMXthough extrapolation.This is valid because RD Xis a much larger problem than isHMX based onits greater solubility in water and usein

explosive components.

80

.E 600,c.-

EE 40s

20

............ ............. ............. .............. ............. ............. .............

.............................................................................................

.............................................................................................

I\...........................................................................................


I ~ T N TPLC e DXHPLCA HMXHPLC

Figure 8 Normalized HPL C forHE with Ozone.

Another reaction wa s performed tocompare the lab results of pure samples to thoseusing actual wastew ater. These reactionsinvolving LAAP pinkw ater were all performed inthe annular reacto rso that the color changescould be more closely monitored visually. TheTNT and RDX we re gone after2 hrs of reactionas shown by HPLC analysis. There were8identifiable byproducts at t he start of thereaction and 6 following 13 hrs of reaction under


9/10

4

ozone. The pH remained basic throughout thereaction at ca. 8.2.

A similar reaction was attempted withTiO, but with important differences. In the firstreaction, a large quantity of inorganic carbon

was m easured (carbonate fi-om hard water). Webelieve tha t this can have a detrimental impacton the m ineralization of theHEs. We also weretaxing the limits of our T OC analyzer with theamount of carbon in solution. Therefore, wediluted the solution volume by half withdeionized water and acidified the solution (with1% HNO, until a pH of 6.0 was reached) undermixing to counteract these problems.

A direct comparison can not now bemade between the two reactions but we plan on

repeating the experiment at zero dilution intitanium dioxide and acidification with halfdilution using ozone treatment.As expected, thenumberof intermediates with th e catalyst andacid treatment was much lower than that withozone (start of7, and ending with2). Thedestruction ofRDX was complete after15minutes and the TN T was gone after1 hr. Theacidificiation removed ca.30% of the carbonpresent initially.

carbon in order to increase the rate ofdestruction ofHEs is viable, however thedilution would have t o be investigated withtoxicological methods before being promoted asa way to increase the rate of reaction. Theproduction of 100 gallons of highly toxic was tefrom 20 gallons of slightIy toxic w asteis asituation that must be avoided.

The use o f acid to remove inorganic

Conclusions

very comparable in photoreaction products andrates. Based on the similarityof the moleculesthis is not surprising butis usefbl since thedevelopment of a technique of removal ofRDXcan be easily applied to one containingHMX aswell. The chemistry ofTNT is much morecomplicated than either of these molecules and is

The chemistry ofHMX and RDX are

primarily related to the greater complexity of themolecule and the presence of aromaticity.

Photremediation can be a viable optioninthe destruction of high explosive contaminatedgroundw ater. The issues centeron cost and

speed o f destruction in either a lamp based o rsolar based operation.

remote areas with high average insolation. Theuse of lamp driven phototreatment whenemploying either ozone or titanium dioxide wassimilar with a few exceptions. The ozonereactions seem to create more intermediates thatmay require longer treatment t o low er thetoxicity to a d ischargeable level but at the sametime lack the mechanical df ic u lt y o f dispersing

and separating a catalyst in a batch process.

Solar photocatalysis can be of usein

Future WorkFuture w ork will involve the

measurement of toxicity/mutagenicityof theproducts of these different oxidation techniques.The use of bioremediation has not beenthoroughly examined, but w e feel this may offerfaster cleanup times if it fo llows initialdestructionof the parent com pounds using thetechniques described herein.

the oxidation reactions based o n productformation. Our labs are in the developmentstages for the de tection of intermediates using ahigh performance liquid chrom atographyinstrument. We are currently obtaining theexperimental UV Nisib le spectra o fintermediates and plan to obtain representativetheoretical spectra of likely candidates usingsemi-empirical quantum mechanical methods.Early results with aqueous T NT are promisingand work with known com pounds will be usedto determine if this will be a viable method, o rwhether another semi-empirical method will givebetter m atches to experimentally obtained data.This will be a case where th e use of a theoreticalquantum chemistry method can be validated at

We also are looking at the mechanism of


10/10

I

the same time that it is applied to a knownchemical problem.

FundingThis work wa s fbnded under DOE

contra ct number DE-AC04-94AL850000.

References1. Prairie, M.R.; Evans, L.R .; Stange, B.M .;Martinez, S.L.Environ. Sci. Technol.,27, 1776-82, 1993.

2. Prairie, M .R .; Stange, B.M.Photoca talysis or the Treatment of WasteWater: Application s Involving the RemovalofMetals, Proceedings from the AnnualAIChEHeat Transfer Conference, Symposium onInnovative Applications of Solar Energy,Atlanta, August8- 11, 1993.

3 . Show alter, S.K.; Prairie, M.R .; Stange,B.R.; Rodacy, P.J .; Leslie, P.K .Photocatalysisfor the Destruction of Aqueous High Explosives,Presented at Emerging Technologies inHazardous Waste M anagementVI, Session No .120, Atlanta, September 19-21, 1994.

4.

P.J.;

Leslie, P.K.; Huang, M .; Datye, A.K.Applications of Photoca talysis to Actual Waste:Explosives D eg rah tio n and Silver Recovery,Presented a t The First International Conferenceon Advanced Oxidation Technologies for Waterand Air Remediation, London, Ontario, June 25-30, 1994.

5. Pal, B.C .; Ryon,M.G. DatabaseAssessment of Pollu tion Control in the MilitaryExplosives and Propellants Production Industry,Oak Ridge National Laboratory Report ORNL-6202, February 1986 and references therein.6. Oriel Light Sources Catalog, page 83 fora lkW att Xe-Hg lamp, catalog# 6293, 1989.

7. Technical Bulletin DRS 080592 GERfrom Light Sources, Inc. for catalog# 6293,1989.

Prairie, M.R.; Stange, B.M .; Rodacy,

8. Wolfram, E. Personal Comm unication.

9. Kaplan, L.A.;Burlinson, N .E. andSitzmann, M.E.,Photochemistry of TNT:Investigation of the Pink WaterProblem,Part II, NSWC /WOL/TR 75-152, 1975.

10. Mohliner, D.M .; Adams, R.N .;Argersinger, W.J. Jr.J . Amer. Chem.SOC.,3618,1962.

11. Sault, Allen G. Unpublished Results.

Annular Reactor

Documents

Transcript of Annular Reactor