(0CONTRACT REPORT BRL-CR-606

EXPLORATORY DEVELOPMENT ON A NEW PROCESSTO PRODUCE IMPROVED RDX CRYSTALS:

SUPERCRITICAL FLUID ANTI-SOLVENT RECRYSTALLIZATION

~ '~,V. J. KRUKONISry M.P. COFFEY,• c c P.M. GALLAGHER

'••. MAY 1 G 1989

sJL ic zq8ao q

APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED.

U.S. ARMY LABORATORY COMMAND

BALLISTIC RESEARCH LABORATORY

ABERDEEN PROVING GROUND, MARYLAND

Reproduced From ,Best Available Copy

[ECURJTY QSI,��;IATION OF THIS PAGE

Form ApprcveREPORT DOCUMENTATION PAGE OBN.00 810. REPORT SECURITY CLASSIFICATION lb. RES.RICTIVE MARKINGSUNClASSIFIED

2a. SECURITY CLASSIFICATION AUTHORITY 3. OiSTRI3UTION/AVAJLAB!UTY OF REPORT

2b. DECLASSIFICATION I DOWNGRADING SC ..DULE APPROVED FOR PUBLIC RELEASE; DISTRIBUTIONUNLIMITED

4. PERFORMING ORGANIZATION REPORT NUM8ER(S) 5. MONITORING ORGANIZATION REPORT NUMBER(S)

BRL-CR-66..

64. NAME OF PERFORMING ORGANIZATION 6b. OFFICE SYMBOL 7a. NAME OF MONITORING ORGANIZATION

Phasex Corporation ( Ballistic Research Laboratory

6C. ADDRESS (Oty, State, and ZIP Code) 7b. ADDRESS (City, State, and ZIP Code)360 Merrimack Street A • •Lawrence, MA 01843 Aberdeen Proving Ground, MD 21005-5066

841. NAME OF FUNDINGISPONSORING 8b. OFFICE SYMBOL 9. PROCUIPEMENT INSTRUMENT IDENTIFICATION NUMBERORGANIZATION (If pgikab, :-

Ballistir Research Laboratory SLCBR-TB-EESc. ADDRESS (Cfty, State, and ZIP Code) 10. SOURCE OF FUNDING NUMBERS

PROGRAM IPROJECT TASK LWORK UNIT '

Aberdeen Proving Ground, MD 21005-5066 ELEMENT NO. INO. I NO. CCESSION NO.

11. TITLE (icnude Sun• y Uasuihcanon) I . •

EXPLORATORY DEVELOPMENT ON A NEW PROCESS TO PRODUCE IMPROVED RDX CRYSTALS:SUPERCRITICAL FLUID ANTI-SOLVENT RECRYSTALLIZATIO -

12. PERSONAL AUTHOR(S) a gh',Krukonis, V. J.; Coffey, M. P.; Gallagher M.. .

13a. TYPE OF REPORT 13b. TIME COVERED / 14. DA TE OF REPORT (Year,,Month.a 1a . PAGE COUNTFinal IOMSep 86 1o 88 88/15/2

16. SUPPLE',AIENTARY NOTATION

17. COSATI CODES 15. S •ICT TERMS (Continue on rlv.,s4 it ncestary and identfIy by block number)FIEqD GROUP SUB-GROUP explosivesj recrystallization"

19 01 RDX) supercritical fluids19 0 / O1 HMX fine particals • "A•

19. ABS TACT (Coninue on ,*rerw it necetuary and •i•entby by block number)Topic A86-142 of the FY 1986 DOD SBIR Solicitation sought investigation (,f methods which

could produce improved intragranular cavitygfree RDX (cyclotrime ;hylenetrinitramine)crystals of 100-150 microns size. RDX produced by the current method, recrystallization fromcyclohexanone-water solution by evaporative concentration, contains intragranular cavities;they can interfere with the performance of the explosive.

Phasex Corporation proposed the use of supercritical fluids as anti-solvents to recry-stallize RDX from organic solvent solution. The cnncept is based upon the phenomenon ofabsorption of gas near its critical point by an organic solvent. The absorption of the gaslowers the dissolving power of the organic solvent for RDX resulting in nucleation of RDX.It was suggested that process parameters such as RDX concentration, organic solvent, gaspressure and temperature, rate of introduttion of gas and similar factors could be controlled:to form the desired particles. Based upon early demonstrations of the supercritical fluidanti-solvent recrystallization 'Žabilities, the main objective of the work viz., the

20. DISTRIBUTION /AVAILABIUTY OF ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATIONS.UNCLSIFIED/U NOMITED C3 AME AS RPT. D OTIC USFRS UNCLASSIFIED

22a. NAME OF RESPONSIBLE INDIVIDUAL 22b. TELEPHONE (Include Arva Cod•e) 22c. OFFICE SYMBOL

WARREN W. HILLSTROHM 301-278-6571 SLCBR-TB-EEDO Form 1473. JUN 86 Pre*Wowoeivonsareobsolere. SECURITY CLASSIFICATION CF THIS PAGE 7 /

-..

:-,". ;i ,'i :..', : .'" : }..'.:.' :.: : ,,. ... -., . . . .

19. ABSTRACT cont.

formation of 150 micron RDX crystals free of cavities, was later modified to include theformation of small {20 micron} diameter intragranular cavity-free RDX crystals.

-During Phase I, a wide range of parameter; was investigated. By the end of thework it was shown that the iupercritical fluid anti-solvent process could be. controlled

to produce RDX crystals of narrow particle size distribution; RDX of 150 microns andRDX crystals of less than 20 micron size were formed from cyclohexanone using carbon

dioxide as the anti-solvent. An economic assessment of the process at the

10•00I,000 lbs/yr production level showed that RDX could be recrystallized at total

capital and operating costs of about $0.50 to $1.00/lb.

......... .

V

TABIX OF CONTENTS

Io SUMMARY ........................ . . . . ........... .. . ....... .

II. BACKGROUND ON SUPERCRITICAL FLUIDS S ............................. 2

A. Phenomena ........ .... .......................................... 2

B. Extraction Process Operation ................................... 3

C. Process Applications of SupercriticalFluid Extraction ........................ . .............. .... 10

III. RDX RECRYSTALLIZATICN WITH SUPERCRITICALFLUID ANTI SOLVEM ................... .......... 14

A. Principle of Supercritical FluidAnti-Solvent Precipitation .............................. . .. 14

B. Experimental Methodology and Results ......................... 21

IV. EXTENSION OF SUTERCRITI(rAL FLUID ANTI-SOLVENT

'ECRYSTALLIZATION TO PRODUCTION SCALE OPERATION ...... ....... 54

V. CONCLUSIONS AND RECOMMENDATIONS- ................ ...... ... . ... ..... 59

V7, LITERATURE CITATIONS. ............................................... 60

4A• A u.••.d;'I F•

U')(,,, .. .. ..! [1L[j

Dis ...t

Dist -•. '.,

Topic A86-142 of the FY 1986 DOD SBTI Solicitation sought investigationof methods which could produce improved intragranular cavity-free RDX(cyclotrimethylenetrinitramine) crystals of 100-150 microns size. RDXproduced by the current method, recrystallization from cyclohexanone-watersoltution by evaporative concentration, contains intragranular cavities; theycan interfere with the performance of the explosive.

Phasex Corporation proposed the use of supercritical fluids as anti-solvents to recrystallize RDX from organic solvent solution. The concept isbased upon the phenomenon of absorption of gas near its critical point by anorganic solvent. The absorption of the gas lowers the dissolving power of theorganic solvent for RD! resultin, in nucleation of RDX. It was suggested thatprocess paraneters such as RDX concentration, organic solvent, gas pressureand temperature, rate of introduction of gas and similar factors could becontrolled to form the desired particles. Based upon early demonstrations ofthe supercritical fluid anti-solvent recrystallization capabilities, the mainobjective of the work. viz.. the formation of 153 micron RDX crystals free ofcavities, was later modified to include the formation of small ((20 micron)diameter intragranular cavity-free RDX crystals.

During Phase I a wide range of parameters was investigated. By the endof the work it was shown that the supercritical fluid anti-solvent processcould be controlled to produce RDX crystals of narrow particle sizedistribution; RDX of 150 microns and RDX crystals of less than 20 micron sizewere formed from cyclohexarone uaing carbon dioxide as the anti-solvent. Aneconomic assessment of the process at the 10.000,000 lbs/yr production levelshowed tLat RDX could be recrystallized at total capital and operating costsof about $0.50 to $1.00/lb.

4

1i

II. BACKGROUND ON 37UPERCRITICAL FLUIDS

During the late 1970's and early 1980's supercritical fluids were highlytouted for their advantages in separating or rurifyi-g many materials. Many"misapplications" of supercritical fluids were investigated by research groupsall over the world. There was a reversal in interest when the applications towhich supercritical fluids were first addressed did not result in industriallyattractive separations processes. Therefore. before presenting the results ofthe RDX recrystallization program a discussion of supercritical fluidsolubility phenomena and viable applications is given in order to acquaint thereader with the quite interesting characteristics of these unique solvents.Some of the viable applications developed to a commercial scale, andtherefore, supercritical fluid extraction is no longer just a conceptualprocess or laboratory curiosity that it was connidtred to b'e in the early1980's. Its application to recrystallizing explosives is tichnically andeconomically attractive, and it promises to be a viable iadustrial processingmethod.

A. Phenomena

The phenomenon of solubility in supe,-.ritical fluids was first reported109 years ago by Hannay and Hogarth1 who described their observations of thebehavior of solid materials dissolved in sup,-critical fluids. Theyinvestigated the solubility of several inorg~nic salts. e.g.. cobalt chloride.potassium iodide, and others, in supercritical ethanol and ether, and theirexperiments were carried out in a high pressure glass cell so that they couldobserve the behavior of the solution as conditions of pressure and temperaturewere varied. The most important result of their experimente was the findingthat at conditions above the critical temperature solely a change in pressurecaused a change in the solubility of the solid, or stating their finding in adifferent way. thay observed that solely a change in pressure caused a changein the dissolving power of the solvent.

The pressure dependence of the dissolving power of a supercritical fluidis not limited to inorganic salt solutes. but instead it is general phenomenonexhibited by all supercritical solvent-solid solute solutions and by manysupercritical fluid-liquid solute solutions. 2 Many technical papersdescribing the results of solubility measurements with a number of organiccompounds have developed the thermodynamic framework, and the framework allowssome reasonably goo stimateg of the behavior of many other solute-solventsystems to be made. -' In the references cited above, the solubility ofnaphthalene has been used as the model system for most of the theoreticaldevelopments, and the solubility of naphthalene has been the most widelystudied experimentally. The solubility behavior of naphthalene insupercritical carbon dioxide is similar to that of other solute-supercriticalsolvent binary systems, and this large data base can be used to describe theconcept of supercritical fluid extraction.

Diepen and Scheffer 6 were the first of many recent workers who studiedthe phase behavior of naphthalene in ethylene at high pressure, and this paperwas followed with more studies from the two authors and theirco-workers7.8,9'10 Of historical interest here, Buchner first studied the

2

solubility behavior of naphthalene in carbon dioxide in 190511; an iiterestingsummary of the early (and controversial) period of super-ritical fluidsolubility study is given in Reference 2. Other workers who examined hsolubility and phag behavior of naphjhalene were Tseknanskayq. et alli,13,14,King and &obertson , Najour and King . MacKay and Paulaitisu, McHugh andPaula*:is . Kurnik and Reidl0, Schmitt and Reid 1", Krukonis, et a11 8 . McHugh.et al•. The solubility and phase behavior of naphthalene have been studiedin methane, ethylene. ethane, carbon dioxide, chlorotrifluoromethane, andtrifluoromethane and the last two groups18.19 extended the studies ofnaphthalene solubility using xenon as the solvent. Thus, as was statedearlier, there exists a great deal of inf.•rmation about the solubility ofnaphthalene in a number of gases. Many other solids besides naphthalene havebeen studied for their behavior in superrritical fluid solvents,biphenyl 9 a2 anthracene21

* phenanithrene . benzoic acld' 3 , and stillmany other8I'-.329,21,

As an example of the pressure dependent dissolving power of asupercritical fluid. Fflure 1 gives data on the solubility of naphthalene incarbon dioxide at 450C . As the figure shows, the solubility of naphthalenein carbon diaxide is very small at low pressure, which ia not unexpected sincelow pressure gases are not good solvents, but as the critical pressure ofcarbon dioxide (71 atm) is approached, the solubility of naphthalene starts toincrease; it increases very rapidly as the pressure is raised above thecritical pressure. The solubility values (which were given in units of g/l inthe referenced paper) @-;roach about 10% (w/-*) at pressure levels of 200 atm.Figure 2 assembles a large Iq ount of data on naphthalene solubility from thepreviously referenced works . The curves shown in the figure give theisobaric solubility values of naphthalene in car on dioxide as a function oftemperature tand the dotted line shows the solubility of naphthalene insaturated liquid and in saturated vapor).

As related earlier, all solid compounds (and many liquids) exhibitsimilar solubility behavior in supercritical fluids. To provide partialcorroboration for the statement the solubility behavior -f another system,silica-water, is shown in Figure 3. The pressure and temperature levels arequite different from those of naphthalene-carbon dioxide shown in Figure 2,(because the critical temperature and pressure of the two solvents aredifferent). but the overall appearanci of the isobars is very similar inappearance. For one more example, the solubility behavior of triglycerides(e.g., soy bean oil) in carbon dioxide is given in Figure 4. The similarityof all the figures is evident.

B. Extraction Process O.!ration

Because of the solubility characteristics shown in Figures 1-4. it ispossible to design industrial processes to extract, purify, and/or fractionatematerials based upon changes in pressure on a supercritical fluid solvent; athigh pressure an extraction can be effected, and lowering the pressureelsewhere in the process causes the dissolved minterial to separate from thesolvent which can then be recycled to the extractor. 'Ae process can beoperated as either a batch or a continuous one depending upon the nature ofthe feed and the nature of the extraction. ,.e., whether it be a purification

3

FICU•.E 1

SOLUBILITY OF ORGANICS IN SUPERCRIT!CAL FLUIDS

70

60

so . Naphth3lent in CO 2

cc f45nCi/--

L1

• 40-J

0

30

20

10PCC0 2

0 i %10 150 20Q, 250

PRFSSUnE, ATN4

4

YIGURE 2 -SOLUBILITY OF NAPETAL IN CARBOM DIOXTDE

30.0

10.0 300

0

Zu. I t90I0w

0z

0 /0/ 8/

0.1 / 70 atm

/

0 10 20 30 40 50

TEMPERATURE (Cc)

5

FIGURE 3

SOLUBILITY OF SiO 2 IN WATER

0.38 ,U,,s0.34 S10 2 -H 2 0

o 0.30. 2,•lc'JS0.26 I-t€

w 0.22- ci0

) 0.180 " T 50 BARS -

H0.! 4 0

19! 0.10

0.06 160 . 0

3 PHASE REGION 0

0.02 QUARTZ -LIQUID -GAS CRITICAL END POINT

160 240 320 400 480 530.

TEMPERATURE °C

6

FIG=J 4 - TOL1IBLITf (W7 TRIZCZ.T MXDE ]JCAIMON DIOXIDZ

10.0- ra

S1.0 ý10

z.0

4H0

crzw

o 0.1300

2,000

0.01110 20 30 40 50 60 70 80

TEMPERATURE (0 )

(or topping), a fractionation. or an extraction from a reaction mass. The useof s aupercritical fluid solvent in separat rg materials was first describedin the literature by Todd and Elgin in 1955i9; their phase equilibrium studiesof a large number of solutes with supercritical ethylene led then to suggesttLat industrial separations processes could be readily devised baoed upon thepressure dependence of solubility.

A schematic diagram of a process which uses a supercritical fluid as apressure dependent solvent to extract an organic substance is given in Figure5a. Four basic elements of the process are shown, viz.. an extraction vegasel.a pressure reduction valve. a saparator for collecting the materials that weredissolved in the extractor, and a compressor for recompressing and recyclingfluid. (Ancillary pu, ps.,valving. facilities for fluid mske-up, heatexchangers, and other eqAipment are omitted .from the figure for clarity andease of presentation.) Figure 5b sho-s the Pxtensive data on the solubilityof naphthalene in carbon dioxide previously given In Fi 0 ure 2, and referenceto Figures 5a and )b is made to explain how the supercritical fluid proce soperates.

Some process operating parameters are indicated on two solubility isobarsin Figure 5b; E1 represents conditions in the extractor, e.g.. 300 atm. 55 0 C,and S, the conditions which exist in the separator. 90 atm, 360 C. Theextractor vessel is assumed to be filled with naphthalene in admixrure withanother materi.-I t1'a'., for this discusaion, is assumed to be insoluble incarbon dicmide. Carbon dioxide at conditiors El is passed through theextraction vessel wherein 4 .t dissolves (and extracts) the naphthalene from theinsoluble material. Thc carbon dioxide-naphthlslene solution leaving theextractor is next expanded to 90 atm throug,1h the yressure reduction valve andas indicated by the directed path in Figure 5b. During the preavure reductionstep. naphthalene precipitates from the solution. (A temperature decrease ofabout 19 C occurs during the expansion of the stream, and thus, the pressurereduction path shown in Figure 5b is an oblique one.) The naphthalene thatnucleates and precipitates when the pressure is reduced is collected in theseparator, and the carbon dioxide leaving the separator is recompressed andreturned to the extractor. This recycle process contriues until all thenaphthalene is diasolved and extracted, the directed line segment 1 - 2 andits -everse on the solubility diagram representing approximately the cyclicprocuss.

As an alternative to extraction and separation using pressure reduction(i.e., the path 1 - 2) the process can operate at constant pressure andtemperature changes in a supercritical fluid can be used to effect theMextraction and separation steps. For example. and again starting at 1, thestream leaving the extractor could be passed through a heat exchanger (insteadof a pressure reduction valve) and cooled to. say. 20 0 C as indicated by thedirected portion on the 300 atm isobar. Cooling the stream would result inthe precipitation and collection of naphthalene in the separator at conditiong3. The carbon dioxide lea%iing the separetor could then be passed throughanother heat exch-,ager loc*rt-e downstream of the separator and heated back to55oC before being recycled to the extractor. A fluid recirculating pump isstill required, but it would supply pressure only for frictional and momentumlosses; these are small.

8

IWIGURtB 5a -EXTRACTION PROCESS FIGURE 5b -OPERATING CONDITIONS

TEMPERATURE (CO

30.0 0 10 20 30 40 50

PRESSUREREDUCTION30

3.%

SEPARATOR4 2L

EXTRACTION w .

09

COMPRESSOR5IS

V? 70 atm

46

9

There are regions where the solubility decreases with increasingtemperature, a behavior not usually exhibited by liquid solvents at roomtemperature and advantages of this behavior can also be made in carrying outan extraction and separation. For example, the extraction of naphthalenecould be carried out at, say. 30 0 C and 80 atm starting at point 4 and a10-fold change in solvent disaolving power could be caused by heating thesolution only about 100 C to 40°C; the naphthalene that precipitates would becollected, as before, in the separator. Cooling the carbon dioxide back to30 0 C would raise its solvent power and it could be recycled for furtherextraction. In another variation, carbon dioxide can be employed like"normal" liqtid sol.vents. As stated earlier, the dashed curve shows thesolubility of naphthalene in liquid and in vapor carbon dioxide. Thus, liquidcarbon dioxide at point L can be used to extract naphthalene from the mixtureanc elsewhere in the process the carbon dioxide can be vaporized; thenaphthalene would precipitate in the vaporizer since the vapor at point Vcontains negligible naphthalene. To complete the cycle, the vapor iscondensed and returned to the extractor. Which specific process to use in anyparticular extraction is based upon an evaluation of many factors. The casechosen for discussion. i.e., naphthalene extraction from another insolublematerial, is an idealized one. In an actual case there may be many materialsin the mixture that exhibit vary'rng degrees of solubility. Determining anoprJmum balance of gas requirements, yield, selectivity, and other factors isdetermined on a case-by-case basis.

It has been expedient to use the naphthalene data to present the conceptof the pressure-dependent solvation power of supercritical fluids, but theprocess concept shown in Figure 5 is quite general, applying to any materialthat exhibits the solubility behavior shown in Figures 2-4.

C. Process Applicationa of Supercritical Fluid Extraction

A large number of applications hen been studied in the laboratory duringthe last ten years; some have progressed to the industrial scale and as anillustration of the many kinds of materials that can be treated bysupercritical fluids a partial list of some of these applications is given inTable I.

Table I

SUPERCRITICAL FLUID APPLICATIONS TESTED

Extraction of Spices. Natural Colorants, Essential Oils,30,31.32

Separation of Ethanol-Water 3 3 ' 3 4 ' 35 ' 3 6

Regeneration of Activated Carbon3 7 ' 3 8

Extr&,tion of Caffeine from Coffee. Tea 3 9 ' 4 0 ' 4 1' 4 2

Extraction of Oil from Soy Beans 4 3 ' 4 4 ' 4 5

Extraction of Vegetable Oil from Potato Chips 4 6

Energy Related Extractions; Coal, Residuum, Oil Shale4 7'4 8'4 9

Extraction of Chemotherapeutic Drugs from Plant Materials 5 0' 5 1

Extraction of Naval Stores from Wood Chips 5 2

10

The fact that supercriticsl extraction "works" in the applications listedabove does not necessarily mean that the processes will be viableindustrially. For example, during the late 1970's much effort was directed tothe development of# supercritical fluid extraction process for separatingethanol from watery'. It was suggested that because carbon dioxide couldextract ethanol from water with less energy than is required in a distillationprocess. supercritical fluid separation would be economically viable. Allfacets of a process must be co:.sidered, however, not just the energy costs.and in the case of alcohol-water separation, the higher capital cost for theequipment outweighed the cost savings of reduced energy requirements, and theprocess was found to be not economically viable. On the other hand hopsextraction using supercritical carbon dioxide is commercial in the U.S.. and alarge plant is being operated by Pfizer Inc. in Sidney. NE". Additionally,General Foods has constructed a coffee decaffeination plant in Houston. TXthat is scheduled for start up in the spring of 198855. The added value tothe improvied products that derive from supercritical extraction more thancompensates for the cost of extraction relative to conventional solventextraction, for example, with methylene chloride. In West Germany, there arethree large supercritical fluid extraction plants in operation, onadecaffeinating coffee at a production rate of 60.000.000 lbs/y 5 , and twoplants extracting hops.

During the past five years many "newer" applications of supercriticalfluid extraction have been examined; examples gf such operations are reactivemonomer purification and polymer fractionation 5758, two areas where theproperties of supercritical fluids can be technically and economicallyattractive. Additionally, recrystallization of materials dissolved insupercritical fluids was first suggested by Phasex Corporation in 19 8 1 60a. butactually tht concept is based upon the principles first discovered by Hannayand Hogarth who were referenced earlier. It is of interest to quote from theclosing statements of their 1879 publication.

"We have, then, the phenomenon of a solid with no measurable vaporpressure, dissolving in a gas

"When the solid is precipitated by suddenly reducing the pressure,... it may be brought down as a 'snow' in the gas ... "

Thus, the supercritical fluid extraction process that was describedpreviously in Section II can be operated so as to produce recrystallizedmaterials. For example, the naphthalene that waa precipitated after it wasdissolved and extracted in the process shown in Figure 5 would assuredly be ofdifferent particle size than the naphthalene in the original mixture. Thus,any soluble solid material can be charged to an extraction vessel, dissolvedin a super-ritical fluid, the solution expanded to some lower pressure leveland the particles formed collected in a separator. Some restlts of using thesupercritical fluid nucleation concept were renorted in 19 83 60b and 198400c;examples of nucleating a wide variety of matei-ials such as steroids, dyes. andpolymers were presented. Figure 6, for example, compares as received andsupercritical fluid nucleated particles of an estrogen; the recrystallizedparticles are seen to be quite uniform and of the order of one micron in size.whereas the control (ra received) material exhibits a quite wide particle size

11

FIGURE 6

Photomicrographs: " "" , "(a) virgin particles of - . 'estradiol; (b) particles of ,. -O-estradiol nucleated , * • • . . -• ' '. " ' : '." ... • " . '•" . '

from an SCF solution ... " ' • 'I

4 . o .. .-. a. . - . .o ' . .

4. • 'o ..

• . . .. ,.• .: :

4L,

Y ,

A -100$

N.2 ) . or ,

•o .

A +'•,,..m • .. .100/6

12

distribution with mat:y large particles present. Othgi groups later 6 eportedon suparcritical fluid nucleation of pharmaceuticals and polymers

For the nucleation process to work as described abuve, the solid materialmust, of coarge, exhibit same level of solubility in the supercritical fluid;the "snow" described by Hannay and Hogarth is formed when a solution of thesolute and supercritical fluid is expanded to a lower pressure.

If a material does not dissolve in a supercritical fluid, the concept asJescribed cannot be employed. RDX does not dissolve in any of the "simple"gases. e.g., carbon dioxide and the light hydrocarbons, methane, ethane.ethylene, and the chlorofluorocarbons. yet a supercritical fluid ckn still beused as a nucleating solvent for RDX. The principle of precipitation in thiscase is different, however, but there are no limitations to the use of theprocess for any material provided that certain solubility relationships existamong the material, the organic solvent, and the supercritical fluid. Thenext section explains the principle of gas anti-solvent nucleation in thosecases where the solute does not dissolve in a supercritical fluid, and theresults obtained on the Phase I program are presented.

13

III. RDX REZQTSTALLIZATION WITH SUFXCRITICAL FLUID ANTI-SOLVETMS

A. Principle of Suptrcritical Fluid Anti-Solvent Precipitnt4 .on

The most widely known supercritical fluid nucleation process wasdescribed in the previous section. the requirement for that concept to beoperable being that the solid material dissolve in the fluid. If the compounddoes not dissolve in the supercritical fluid, a supercritical fluid cannonetheless be employed in a process to recrystallize the material, viz., itcan act as an anti-solvent the general requirement being that the material tobe crystallized dissolve in some organic solvent and that the fluid alsodissolve in that solvent.

Results using supercritical fluids to segarate solutions of liquids werefirst reported in 1959 by Elgin and Weinstock 3. The researchers showed thatsolutions of water and an organic solvent could be separated into a water-richphase and an organic-rich phase with suprcritical ethylene, and they termedthe process "supercritical fluid salting out". The salting out process isquite generally applicable to water-organic solvent solutions if thesupercritical fluid is miscible with the organic liquid, if the concentrationof organic liquid in the solution is "high" (with "high" a function of thespecific organic liquid). and if the intermolecular forces between water andthe organic are not "too great". For example, the workers showed that asolution of water-met.yl ethyl ketone (MEK) can be split into a water-richphase ond an organic-rich phase by the action of supercritical ethylene. Inexplanation of the process of separation, the supercritical fluid, ethylene.which is miscible with neat MEK. will dissolve in the water-MEK solution; whensufficient ethylene has dissolved, it will act as an anti-solvent for water,and two phases will be formed.

More recently Paulaitis. et a164 report on the phase equilibria ofisopropinol water-carbon dicride solutions, and it is instructive co reproducetheir data here as an aid in explaining the events that occur when carbondioxide is introduced into a solution of, say, 50/50 (mol %) isopropanol-water. Figure 7 shows the ternary diagram for isopropAnol-water-carbondioxide (at conditions of 136 atm and 60 0C). "Starting" at the 20% IPA/80%H2 0 concentration point (point 1), if carbon dioxid is introduced (depictedby the directed dotted line) and mixed to equilibrium with the aolution, thecarbon dioxide will dissolve into the solution. When "sufficient" carbondioxide has been dissolved, the miscibility characterictics of isopropanol andwater will be modified and a second phase will appear (point 2); the amount ofthe two phases and the concentrations in each phase will be governed bythermodynamic equilibrium considerations for the three components at theconditions of temperature and pressure. The specific concentrations of thephases can be obtained from knowledge of the "tie lines", i.e., theconcentrations of the two equilibrium phases at any combination of components,temperature, and pressure, and the amounts of each phase can be obtained fromthe "lever rule", the material balance relating the "urnixed" composition tothe composition of the equilibrium phases. For example, if enough carbondioxide is added to the solution to reach point 3, the two equilibrium phaseswill have concentrations cf 55% IPA (45% H2 0) for the isopropanol-rich (carbon

14

/

WVIJM 7

Measured Equilibrium Tie Lines for Isopropanol-Water-C0 2

I PA

T 60a CP a 2000 psi

H2 0 CO 2

&/

15

dioxide fret basis) phase, and 10% IPA (90% H2 0) for the vatei-rich phase.These concentrations are determined by the tie line drawn thrcugh point 3intersecting the two-phase envelope at the equilibrium concentrations.

A similar concept was later gppliad to tae separption of polymer-organicsolvent solutions, by Irani et al and McHugh. et al" who showed that thelower critical solution temperature (LCST) of the polymer solution could belowered dramatically by the injection of a supercritical fluid. They showedthat the separation process could, thus, be carried out at 4uite modeattemperature which can reduce the overall energy requirements and additionallycan reduce thermal degradation of the polymer. T-he explanation for polymerprecipitation from organic liquid solution is s:milar to that for theseparation of water from the organic solvent discuzsed earlier. The organicliquid is a good solvent for the polymer, and the supercritical fluid, say.carbon dioxide again, is not. The carbon dioxide dissolves in the solution.the solution expands to the point whereat the carbon dioxide-organic phase isno longer a solvent for the polymer, and the polymer precipitates. Inaddition to solubility effects, such considerations as the molecular weightand molecular weight distribution, and polymer composition, influence theextent of precipitation and recovery of polymer from solution.

Polymers can be separated from solution by solely thermal means, and anunderstanding of the thermal separation concept can be obtained from a phasediagram for a "general" polymer-organic solvent solution. Figure 8 shows apart of a P-T (pressure-temperature) projection of a three dimensional(pressure-temperatu-,e-concentration) diagram; this is a two component phasediagram, viz., polymer and organic solvent with no supercritical fluidcomponent present. (In actuality Figure 8 is a "pseudo-two" component diagrambecause a polymer usually consists of a quite wide range of oligomers which.although they are all homologous series members, influence the thermodynamiccharacteristics of the system smewhat.) The designation L connotes a singleliquid phase region. LL a two liquid phase region, and LLV two liquids and avapor region. The L region is where the solvent and polymer are miscible.

The diagram points out that one method of separating a polymer from itssolution in by raising the temperature of the solution isobarically (asdepicted by the Path 1 - 2); the solution separates into two phases which canbe separated, in a decanter for example. As will be discussed later, thetemperature levels in the region of the LCST line are usually quite high. andthus crossing the LCST line by raising the temperature as depicted by Pathr (?)1 - 2 may not ")i & viable method industrially if the polymer can degrade athigh temperutare. (Incidentally, the two phase region in Figure 8. designatedLL, consists of tNo liquid phases rather than a liquid and a solid polymer(LS) because the temperature levels are frequently near or above the softeningor melting point of the polymer. Additionally. even if the temperature isbelow the softening point there still might be two liquid phases inequilibrium because noncrystalline polymers can be swelled and plasticized bysolvents, i.e., the solvent can dissolve in the polymer.)

The physical explanation for the split separation of polymer when thesolution is heated resides in thermal expansivity considerations. If a liquidis heated to near its critical temperature it expands considerably, and

16/

PHASE DIAGRAm. (P-T PROJECTICV) OF POLYeRZR - ORGANIC SOLVEVT SYSTEM!

LN

* "

L LL

m LL.Va.

L LV

TEMPERATURE

17

17\

N\

---------------------- ?

its ability to dissolve polymers (and other substances) is decreased markedly.Thus if a solution ccntaining dissolved polymer at a high conzer.tration isheated, part (and often e substantial amount) of the polymer will precipitate.Unfortunately, as was said above, the criticel temperature -f a "good" polymersolvent is usually quite high, e.g., 200-3000 C, and thus thermal degradationcan ensue with some polymers and, furthermore, f-tom an industrial processingstandpoint the energy requirenents for carrying cut the siparation can bequite high.

The LCST line and thus separation of polymer from solution can be shiftedto much lower temperatures by the expedient of dissolving a supercriticalfluid in the polymer-solvent mixture; for example, McHugh et al report thatdissolving ethylene in a system composed of hexane-polyethylene co-polypropylene can shift the LCST curve to a regime of much lower temperature.Figure 9 (from Reference 66) shows this. The three P-T projections in Figure9 show the repositioning of the LCST line toward decreasing temperature withincreasing ethylene concentration. A change of 50 0 C occurs with an ethyleneloading change of only 10.1 wt%, from 9.9 to 20 wt%. Incidentally, theposition of the LCST line with solely hexane is off scale in Figure 9, 100+0 Chigher than the 20% line. Thuis, the precipitation temperature can be reducedmarkedly with the use of a supercritical fluid, and thus is especiallyadvantageous for heat sensitive polymers.

The same phenomenon, viz., the addition f a supercritical fluid to asolution of a crystalline solid and an organic solvent, is applicable toseparating/recrystallizing a solid expiosive providing that the explosive isnot "too" soluble in the supercritical fluid. Considering now RDXspecifically, if a saturated, or partially saturated, solution of RDX (incyclohexanone. for example) is contacted with a supereritical fluid, the fluidwill dissolve in the solution, will lower tl,.e dissolving power of thecyclohexanone for the RDX, and the RDX will nucleate and crystallize. Theresultant particle size and size distribution will be a function of manyparameters which are explained in this section.

In many respects the supercritical fluid anti-solvent process isanalogous to any liquid anti-solvent process that operates industrially. Forexample. production of Class 5 (25 micron particle size) RDX is one suchindustrial process. In its operation an acetone solution of RDX is admixedwith an anti-solvent, water, which results in precipitation of RDX. Theresultant mixture of particles and liquid is filtered to separate theparticles of RDX, and the water-acetone filtrate is separated by distillationto rQcover the acetone for reuse. Another solvent, cyclohexanone. is alsoused in the RDX manufacturing process, specifically in .he production of Class1 RDX (100-150 microns). In the recrystallization step a solution of RDX incyclohexanone-water which has been seeded with RDX crystals is heated toevaporate the liquid; with "proper" agitation RDX nucleates and grows on theseed crystals, the specific combination of such parameters as agitation.surfactant, and evaporation rate resulting in the formation of the desired100-150 micron particles. The evaporated liquid (and water) is subsequentlyseparated by di.'tillation, and the cyclohexanone is recycled.

In the process where a supercritical fluid would be used as the

18

FIGURE 9

PHASE DIAGRAM, (P-T PROJECTION) OF POLYMER -HEXAN•E - SUPERCRiTICAL ETHYLENE SYSTEM

130 -"

120 ETHYLENE LOADING

100 L LL

90E 80 90 .. %.9

70 LSL LL.,

cr 60 - LL50 -

L 40-

30

20 LV LV

I10 -,

I00 g ,, 3 I

30 50 70 90 110 130 150

TEMPERATURE (0C)

19

-- -I

anti-solvent, no distillation step is required. The supercritical fluid-organic solvnnt solution that leaves the filter can be separated into itscomponents by a simple pressure decrease which causes the gas to be releasedfrom snlution, and both the gas and the liquid solvent cAn be re--used. Thechoice of which process to use where is, like for any industrial process.governed by many factors some of which were addressed in Section II; there aremany advantages of supercritical fluid anti-solvent precipitation of RD". andthey are developed after the results of the laboratory program are presented.Briefly, here, the separation of solvent and anti-solvent can be made withquite low ener"7 requirements. and it is discussed subsequently that therecrystallization period can be reduced to a duration of literally minutes orseconds. Additionally it is speculated later that the process can be modifiedso as to be able to coat the recrystall'.zed explosive with polymers (or othermaterials) so that they can be stabilized against hydrolytic attack andincorporated into a polymer matrix with ease.

2LI

B. RzparIental Nethodolor

In outline of the experimental program that was carried out. acetone andcyclohexanone were used as organic solvents for RDX, and combinations ofpressure and temperature. and other supercritical fluid par.--eters that wouldrecrystallize RDX from the organic solution were tested. Other parameterssuch as the rate of introduction of the fluid into solution and initial RDXconcentration were also considered to be possible influences on the particlesize, shape, and particle size distribution of recrystallized RDX and werestuded. The supercritical fluids tested were carbon dioxide, ethylene, andethane.

Two er-ierimental systems were used in the studies, one a sight glassassembly which allows the recrystallization process to be observed, and theother a closed vessel assembly. A schematic diagram of the sight glassassembly, a standard 5000 psi "Jerguson gauge" (Jerguson Gage and ValveCcapany). is shown in Figure 10a; a photograph of the assembly is shown inFigure 10b. The photograph is centered primarily on the windowed cavity ofthe sight glass, but the schematic diagram sht es all the features for carxyingout and observing a recrys,.allization test and collecting sample forsubsequent examination. A thermocouple is positioned in the solution tomeasure the temperature, it is part of a temperature control circuit not shownin Figure 10a or 10b. A pressure gauge for measuring pressure is positionedat the top of the assembly. The variour valves shown permit gas to beintroduced, liquid to be drained, and/or gas to be removed from the solution.The cotton plug at the bottom of the cavity zerves as a filter for thecollection of particles of RDX when the expanded liquid is drained.Precautions were taken to avoid direct contact of sharp metal surfaces withthe RDX crystals.

In explanation of the conduct of a recrystallization experiment, anamount of MDX-cyclohexanone solution is charged to the sight glass and thesystem sealed. Carbon dioxide is admitted at the bottom, and the bubbles ofgas entering the cavity both mix the liquid and dissolve in it. As wasexplained earlier, as the pressure is raised, the carbon dioxide increasinglydissolves in the RDX-cyclohexanone solution, and the solution expands. At"come" pressure level enough dissolution of gas and concomitant expansion hasoccurred to decrease the dissolving power of cyclohexanone for RDX, at thispoint crystals begin to nucleate and grow. (A linear motion mixer had beenplanned for use for the recrystallization studies but because of the knife-edge metal surfaces of the agitator (which colld initiate decomposition ofRrX), it was not used during the recrystallization tests. During the neatsolvent expansion tests the mixer was used and the results ')f the expansion-pressure experiments were found to be about the same whether the mixer wasenergized or not. indicating that the agitation caused by the bubbles ofaupercritical fluid passing through the solvent provided reasonably extensivemixing action and that there was little or no resistance to dissolution of thegas.)

It is difficult to predict a priori the particle si'e, and shape, andparticle size distribution that will result from some speecit:!c combinations of

21

FiuelOu &aheuatio Diagram of Sighit Glaza I- - Presunre Gauge

Figure l0b Pbotograph at Sight 01asa Assembl.

Vff

f7!

Thermocouple

71 Cotton Plug

Gas Inlett

ftPatrded Solution Outlet

22

temperature, prescure. and concentration parameters, and the goals of theresearch were to determine these relationships. It is of value, however, tosummarize soue simple recrystallization theory insofar as the theory guidedthe experimental program in its early stages.

The equilibrium in assembly of dissolvted molecules which can combine toform a critical nucleus beyond which size favorable fluctuatio-Is will causethe nuclei to grow has been described by Gibbs; he presented the 6 onditions ofcritical nuclei formation based upon fr.e energy consideraticns. , andAdamson 8 gives an excellent treatment of Gibbs' mathematical description.Since the formation of nuclei requires tho formation of an interf&ce betweentwo phases, the free energy of tb,' systen %ill usually increase initialljy,Until an embryo (a particle smaller than a stable nucleus) reaches somecritical diameter its growth demands that the interfacial energy incrsase.Once the particle is of sufficient size, i.e.. of critical diameter, there aretwo competing modes of lowering the particle free energy: the nucleus can growindefinitely, or it can shrink and disappear. It should also be consideredthat critical nuclei grow at the expense of subcritical embryos so that thefinal particle size is dependent upon the initial nuclci 2oncentration.Adamson also presents a readable development gf the rate equations describingnucleation first derived by Beckar and Doring 8 Usiug the free energyconsiderations Becker and Doring dariveL an euation fo;- the rate of formationof nuclei of this critical size. viz.,

rate = Ze-FaIRT (1)

Z is the collision frequency (calculable from classical kinetic theory)

Amax is the Gibbs expression which derives from nucleus surface energyconsiderations

The free energy relation is given by

AFmax= B/[RTln 8]2 (2)

B is a constant derivable from physical properties of the system.

s is the supersaturation ratio. (For purposes of illustration, in avapor-liquid system a is the ratio of actual component pressure to thenormal vapor pressure; in a system involving a solute dissolved in asolvent. s is the ratio of actual concentration to saturationconcentration.)

From Eqs. 1 and 2 it is seen that the nucleation rate is very strongly'nfluenced by the supersaturation ratio. As an example of the rapid increase.iJ the rate, Adamson shows for the condensation of water vapor that the numberof nuclei formed (per second per cc for water vapor at 0°C) is only 10-8 at asupersaturation ratio of 3.5 whereas it is 103 at a ratio of 4.5. The ratio

23

at which the nucleation rate changes from iwperceptible (e.g., 10-8) to verylarge (e.g.. 10 ) is termed the critical* supersaturation ratio. The rate ofnucleation at that point is termed catastrophic, and the phenomenon,catastrophic nucleation. That "some" level of supersaturation is aecessarybefore embryo formation will occur is found in the observations ofsupercooling of polymer solutions, the requirement of the scratching ofbeakers to promote precipitation, seeding of clouds with silver iodide topromote rain (which is first formed as snow), etc.

Continuing with the quite general description of the total particleformation process, the size of the final particles achieved in any particularsituation is governed by the rate of formation of critical nuclei and by therate of growth of these nuclei. The kinetics of nuclei formation involvesboth free energy and rate of molecule transport at the interfacial boundary.the latter being especially difficult to describe in quantitative terms. Theoverall rate of growth of the nuclei once they have been formed during thecatastrophic period can be given by the mass transfer relation

flux of material to surface = kAAC (')

k is a mass transfer coefficient

A is the surface area of the particle at any particular instant

AC is the concentration driving force at any particular instant; isgiven by C-Ce. (or in words, the concentration of the component .n thegas minus itsequilibrium concentration.)

Thus, particle size and particln size distribution (PSD) is determined bythe interaction between the nucleation rate and the growth rate of crystals.on one hand, and by the rate of creation of supersaturation. on the otherhand. The interaction can be demonstrated by the schematic representation ofcompeting events in Figure 11. in which supersaturation s is plotted againlsttime. If supersaturation is created by the addition of an anti-solvent, therate of addition will affect the rate of supersaturation creation, and it willbe. of course, represented by the slope of the line in the graph.

There are three regions designated as the 1lorizontal sections I. II. andIII depicted in Figure 11. In region I tLe solution is undersaturated, i.e..the actual concentration is less than the thermodynamic equilibriumconcentration, and no nucleation is taking place. In region II the solutionis supersaturated but the supersaturation value is below the so-calledcritical supersaturation: no nucleation is observed, but any existing crystalsin the mixture will grow. In region III both nucleation and growth are takingplace.

"The word critical here is not related to the critical point of a gas orliquid.

24

-. /

VIGMR 11

msUkmmTUATw vs TnhE ni PRCIPITATION BY COwnTIousADDITON OF hArrI-SOLVMfT

* SU~tRSAU7201O vs TIME IN PERCIPITATION B1 CONTINUOUSADDITION OF ANTI-SCLMBT

S CriticalSupersaturation

S 0

(n

Time

25

The size or the PSD of the precipitated crystals will depend on thenucleation rate and growth rate of the crystals (which are determined by thesupersaturation. temperature, pressure, etc.) and the rate of supersaturationcrestion which is determined by the rate of anti-solvent addition, theoriginal solute concentration in the solvent, and other factors. In otherwords, the particle size and PSD will depend upon the balance between theformer processes which "consume" supersaturation and the latter, thegeneration of supersaturation.

Thus, in curve A the generation of supersaturation predominates over theconsumption, and the system nucleates continuously (continuous polydispersePSD). curve B represents cases with periodic nucleation events (discrete PSD),and in curve C nucleation takes place only in the beginning of theprecipitation followed by growth (monodisperse or very narrow PSD). If incase C nucleation had been made to occur just above the criticalsupersaturation, a small number of nuclei would most probably form andconsequently the final product would proba'ly consist of rather large crystalsof nearly uniform size which is the condition for the formation of largemonodisperse crystals.

The rate of creation of supersaturation is related to the rate ofexpannion of the RDX-organic solution. The expansion of neat cyclohexanone.i.e.. with no RDX in solution, and neat acetone was first determined in thesight glass. (The expansion was determined to guide subsequent "blind" testscarried out in the closed vessel assembly.) The volumetric expansion curvesof cyclohexanone and acetone by carbon dioxide are shown in Figures 12 and 13.As the figures show, the volumetric expansion is quite large, and thisphenomenon of absorption of carbon dioxide into these solvents and concomitantexpansion is. of course, the basis for the supercritical fluid anti-solventconcept that was proposed and studied.

It is most facile and instructive to describe by means of photographs theexpansion events that are observed in the sight glass during the introductionof supercritical fluid. The photographs will also give an indication of howthe expansion curves in Figures 12 and 13 were obtained. Figure 14 is aphotograph of the sight glass assembly, and the initial level of cyclohexanoneis indicated by the arrow; a volume of 15 ml of cyclohexanone bas been addedto the cavity in the sight glass whose internal volume is 60 ml. Uponintroduction of carbon dioxide the solution begins to expand and the per centexpansion at any pressure level is obtained from the observed level change andthe volumetric calibration of the sight glass cavity. Figure 15 shows theexpansion that has occurred at an arbitrary pressure level, 700 psig at 20 0 C,and the percent expansion can be calculated (and plotted as one point inFigure 12).

In order to determine recrystallization behavior and its relation to suchfactors as the extent of expansion at the onset of nucleation, etc.. the sightglass assembly is again used. A cyclohexanone solution of RDX is contactedwith carbon dioxide at increasing pressure in the same way as was the neatcyclohexanone, and the events that occur, e.g.. nucleation as a function ofextent of expanpion are observed visually. It is again instructive to present

267/

FIGME 12 -VOLUMMTRIC EXPANSION 0F

CTrML0BhAIONE BY ChJRN DIOXIDE

91C 0

- Ic

27

I'IG'TI 13 -VOyLM(ZTRC EZPMNIONC 01 AC=T?M

BYCURIO DICYID

*S' 25 c

/6'c 4o'00c

A.8

77 )= TICURE 14 -SIGHT GLAS SHOWIING

INITIAL LEVE OF CYCLOEEANONE

FIGUR 15 -SIGHT GLASS SHEK)ING

LEVE OF CYCLOHEWINONE AND CARBON

r - Lquid evelDIOXIDE AT 700 PSIG

29

the events that are observed by means of photographs. Figure 16 is aphotograph of the sight glass that has been charged with 15 al of 10 RDX-cyclohexanone solution; the initial level of the solution is indicated and theaxial mixing action was not energized for this test. As carbon dioxide isadmitted, the solution expands as the gas dissolves in th; organic liquid. Atsone sufficiently high extent of expansion minute crysta._ are seen to form.Figure 17 depicts this condition, and the le-;el of the solution .* indicated.The solution has become hazy because of the presence of small particles ofRDX. and this condition is termed "onset of nucleation". Fo- strict accuracy,here. the turbidity of Figure 17 is "slightly" beyond the true onset so that amore visible condition could be photographed. At the true one@t ofcatostrophic nucleation that is governed by the thernodynamic and kineticfactors discussed earlier, the actual nuclei themselves are not visible sincein theory they comprise assemblies of "a few" thousand molecules. Theappearance of the visible haze nonetheless was a convenient measure in thelaboratory of the effects of changes in parameters.

Digressing for the time being from further discuEsion of therecrystallization experiment whose start was shown in Figures 16 and 17. it isof value to discuss the "onset of nucleation" condition further here in orderto present how it is related to the extent of volumetric expansion and howstill other factors such as the rate of introduction of gas can influence boththis onset condition and the size and shape of the crystals ultimately formed.In the particular experiment whose photographs are shown in Figures 16 and 17,the gas pressure level of 700 was reached within a few seconds because of theparticular rate of introduction of the gas. If on the other hand, the gas isintroduced at a much lower rate. say, over a period of the order of a fewminutes. the onset of nucleation is seen to occur at about 100 psi or morelower; additionally, during the slow introduction it is seen that the degreeof turbidity is not so great at onset which is an indication that fewerparticles are being formed at that particular time. In fact. in the slowaddition case the number of particles that are seen to form to a size to bevisible are so few as to almost defy detection; they are identified to bepresent in the liquid by the expedient of moving the backlighting to variouslevels and angles and visually determining the light reflected from thecrystals.

In brief summary here. then. even the "relatively simple" observation ofthe onset of nucleation is in some way related to a pressure-time function.The number of crystals observed to form is also related to pressure-time, andassuredly so is the size and shape of the crystals. The phenomenon is quitecomplex, but size-shape phenomena can be related experimentally to a number ofparameters which can be selected and controlled, and results ofphotoicrographic analysis are presented subsequently.

Returning. now. to the discussion of the recrystallization experimentshown in Figures 16 and 17, to complete the test to precipitate all (or almostall) the RDX that was in solution the pressure is raised still further causingthe solution to expand still further, causing in turn a still further decreasein the solvent power of the cyclohexanone phase for RDX which, of course.results in both crystal growth and additional nucleation. The re-.ults ofreaching a pressure level of 850 psig is shown in Figure 18. The

30

( FIGURE 16 -SIGHT GLASS AMUInITIAL LEME OF RDX--CYCLOMIXAMNOE

Liqui U"ISOLUTION

~ f~ FIGURE 17 -SIGHT GLASS SHOWING

-'~-~-- -~ONSET OF NUCLEATION AT 700 PSIG

CARBON DIOXIE

Liquid Le*vel. FIGURE 18s SIGHT GLASS SHOWINGFULL MANSION AND SETTL.ING OF

RDI GIYSMAS

31

cyclohexanone has expanded almost two-fold and most of the RDX crystals havesettled and have collected at the bottom of the sight glass cavity. To obtaina sample for micrographic analysis the cyclohexanone-carbon dioxide phase isdrained through the port below the cotton plug. and the crystals collect onthe cotton. The crystals are "washed" once (or more times) by the expedientof introducing more gas to its liquefaction point, and the adheringcyclohexanone is dissolved. Draining the carbon dioxide (which now containsthe previously adhering cyclohexanone) results in virtually dry RDX crystalswhich are removed from the sight glass by means of a long handled spatula.

The sample of recrystallized RDX is prepared for microscopy evaluation byplacing a very small amount of the RDX on a drop of toluene that had beenplaced on a glass slide'with a fine tip dropper, dispersing the powder anddrop of toluene. and spreading the dispersion on the glass slide to ensurethat the particles are distributed and a cover slide placed on the dispersion.The slide is dried (to evaporate the toluene). a drop of a dispersion oil isdistributed over the particles, and a cover slide placed on this dispersion.The particles are examined with transmitted light. The refractive index ofthe two dispersion oils used was 1.500 and 1.600. respectively; the refractiveindex of RDX crystals -1.598. 1.602. and 1.585. respectively, for the threecrystal faces. The crystals will be somewhat difficult to distinguish in someof the photographs, therefore, but any intragranular inclusions will appear asdark (or opaque) spots. rhotcuicrographs were taken at primarily twomagnifications. 150X and 300X. which convert to scale markers of 15 mm/100microns and 30 mm/100 microns. respectively.

Figures 19a and 19b are photomicrographs of the as supplied RDX producedby Holaten Army Ammunition Plant. (The written notes on the photographs arethe microscopist's designation for test number, file ntmber, magnification.etc.. and are retained for ease of comparison of the magnification and scalemarker.) The average particle size is seen to range from about 10 microns to400 microns, and almost all the crystals contain intragranular cavities.These cavities are seen to be quite large, in some instances, of the order of20-50 microns or more. These cavities are voids that result when the liquidthat has baen trapped in the crystal during the crystal growth processevaporates/diffuses from the crystal; as stated in the Summary, it was theformation of RDX crystals without voids that was the goal of this program. Asrelated in Section I the main objective of the program was to investigate thegas anti-solvent recrystallization process for its potential to form 100 - 150micron cavity-free RDX. During a s.-te visit early in the program byDrs. Warren Hillstrom and Robert Frye and a demonstration of the capabilitiesof the process, especially as to the ease of formation of very smallparticles, less tLan, say. 10 - 20 microns, the program was modified so as toinclude also the investigation of parameters which would form RDX ofcontrolled small size. The discussion of results that follows includes bothaspects. As is shown, the process can be tailored to produce both small andlarge particles of intragranular cavity-free RDX.

Depending upon the factors and parameters mentioned earlier, viz.,pressure, temperature, concentration of RDX, rate of introduction of gas,etc., a variety of shapes and sizes of crystals could be formed by thesupercritical fluid anti-solvent process. As was discussed previously in a

32

fIGURE19. -FIG=R 19b-

XDX CRySTAS PRODUCE AT UOLSTEM hAP. L-ARM~ KAMNXICATION OF fEDI CRYSTALS

OPAQM SPOT ARE INT AGRAIMLAR CAVITIS N&DZ TO ACCENT THE SIZE OF INTRAGRA2IULAR

CAVITIES

4* ie1Ut.4

Jo-/

2C

33

general way, a rapid increase in the superaturation ratio will lead to a highnucleation rate, a large depletion of solute concentration, and concomitantsmall particles. A rapid introduction of gas into the vessel causes rapidexpansion of the solution, and the expansion, of course, lowers the dissolvingpower of the cyclohexanone, which is equivalent to an increase insupersaturation. 'y was shown visually in Figures 16, 17, and 18. if carbondioxide is injecto. to some pressure beyond that "threshold" pressure (THP) atwhich particles fi-it can be seen to appear, the nucleation is massive andoccurs almost iiiscantaneously. depleting the solution in RDX. A:cordingly. inthis cage, the final particle size is small due to the lack of materialavailable for subsequent growth, and tho PSD is quite narrow. Figure 20 showsthe general trend for recrystallization with rapid injection. Importantly,very few of the RDX crystals show evidence of the opaque spots that areindicative of the intragranular cavities.

Threshold pressures (which are near where the expansion curve becomesexponential) at various RDX concentrations were determined by observation inthe sight glass. It was found that when RDX solution concentration isdecraased. the pressure required for onset of nucleation increases, or statingthin result differently, the lower the concentration the higher the extent ofexpansion required to cause RDX to nucleate. Figura 21 is a graph of theextent of expansion required to cause onset of nucleation for RDXconcentration in cyclohexanone. The trrxd is quite easy to explain. At highRDX concentration very little anti-solvent (i.e., very little expansion) isrequired to begin tc precipitate the RDX because a supersaturation conditionis reached quickly as carbon dioxide is injected. At the very lowconcentration end L large amount of expansion is necessary to reach asupersaturation condition. These results are analogous to the events duringthe precipitation that, for example, occurs by cooling of RDX solutions of tworespective concentrations, viz., a high concentration one and a lowconcentration one. If. for example, a solution of RDX in cyclohexanonesaturated at. say 80 0 C. is cooled, a drop in temperature of only a few degreeswill result in the formation of crystals. If a low concentration solution,say. 11, is cooled from an initial temperature of 80 0 C, no. precipitation will ,occur until some very cold temperature probably well below room temperature isroached.

With slow injection of the gas, of the order of 3 to 5 min, nucleationbegins to occur again at the "threshold" pressure (THP); onset of nucleationand THP are used interchangeably in subsequent discussion. The particles thatare formed depend upon the procedure that is followed after the onset ofnucleation at the threshold pressure. If gas injection is terminated at theTHP (and after onset has occurred) and the system is allowed to remain forscme hold time, the nuclei that are formed will grow. Because fewer nuclei\are formed at this condition than in the case of rapid injection far beyondthe threshold pressure the crystals can grow much larger. Figures 22a and 2"bshow that very large crystals can be produced. Figure 22a shows RDXrecrystallized from cyclohexanone and 22b from acetone. Some very smallcavities are seen in the crystals formed from acetone solution, but as seenfrom Figures 20 and 22a and as will be seen subsequently, very fewintragranular cavities are present in RDX recrystallized from cyclohexanone.In the slow injection situation when gas injection is continued beyond THP,

34

fMGUR 20 - TIPICAL RDX CRYSTALS PRODUCED WITH RAPID

INJECTION OF CARBON DIOXIDE INTO RDX -

CfaCLOHE0ANON SOLUTION

I I "

\zt 4" F

'A ..l,

I c

35

FIGURE 21 - VOLUMETRIC UiPASION E(QUTIMED OF RDX--CYaCHEZ•AM"

SOLUTIONS TO CAUSE ON-MET OF NUCLEATION

/0

C40lIC 97A/re Ar/ -0A/,

36

"U ,/,,

FIGURE 22.a - FIGURE 22b -

TYPICAL CRYSTALS PRODUCED FROM RDX- TYPICAL CRYSTALS PRODUCED FROM RDX-

CYCLOmA3NNE SOLUTION WITS SLOW ACETONK SOLUTION Wfl SLOW INJECTION

INJE CTION AND LONG HOLD AT ONSET OF AND LONG HOLD AT ONSET OF NUCLEATION

NUCLEATION

37

V, ,. .. ,

UIGURE 23 - TYPICAL CRYSTALS PRODUCED FROM RDX-,

ACETONE SOLUTION WITH SLOW INJECTION

CONTINUING PAST ONSET OF NUCLEATION

(WZDE PARTICLE SIZE DISTRIBUTION)

a. ,

9 .' . .-

, - - ,

I W.K) .

38

' . ' • " . .-.

the simultaneous occurrence of "new" nucleation and continuing growth producesa very widc distribution of particle sizes. Figure 23 shows RDX crystalsformed during slow carbon dioxide injection beyond THP; they range from tensto hundreds of microns in size.

The generalizations made above are sufficient for describing theresultant particle size and PSD for almost all subsequent tests. In briefsummary. with rapid injection beyond threshold pressure it is understood thatparticles will be fairly small and PSD fairly uniform regardless of hold time.In discussing the various sets of results, the designation rapid (30-60 sec)or slow (3-5 min) gas introduction will be referred to. In some of the slowinjectiou tests in order to ansure that the particles grow as much as possibleand also to provide some semblance of "time control", a hold time of -20 minwas used; this is designated "long" hold time (unless otherwise indicated).

One of the most significant variables to consider is the choice ofsolvent, since the final particle size, PSD. and yield (recovery of RDXcrystals from solution) are influerced by the solubility of RDX in the organicsolvent. Table III shows the temperature-RDX concentration pattern for twosolvents that are used in the manufactur.'ing process at Holsten AAP. (Manyother common solvents such as the paraffins and aromatics exhibit virtually nodissolving power for RDX.)

TABLE III

Solubility of RDX in Organ=i Solvents

TC Solubility (g/lOOS)

Acetone 20 7.3

40 11.5

60 18.0

Cyclohexanone 25 12.7

97 25

Regardless of the operating conditions, there is a startling differencebetween recrystallization of RDX from acetone and from cyclohexanone. Withonly a few exceptions use of cyclohexanone as the solvent produces smallerregular-shaped particles, whereas with acetone the particles were quiteirregular and sometimes hundreds of microns in size at the same externalconditions of temperature, pressure, rate of injection of gas. etc. Thereasons for these differences are not known at present, and these differencesare considered somewhat surprising. Table IV. page 73 from an Aberdeen PGreport (AMCP706-177) is a listing of solvents, RDX solubility. and thecrystalline form of RDX that is recrystallized from the solvents by the"normal" procedures described earlier. The table shows that thick hexagonalcrystals are formed from acetone solution, and thick hexagonal is the form of

39

TABLI V I

SolubiUity a-4 Cryatal Form of 9D1 /X/7.

€_ e (Rn) AMCP 706-177

iubility oft Cyelonite, Holstoo :ot E-2-5 in Various Solvents:

Sol~ubilit

Solvent

Pit Grade orsolvent 8' S ce 280 C tHeated Crystalline Torm

Acetone 5. C 8.2 16.5 at 60°C hexagotal-thickc clchefLnone 155.6 CP 13.0 2,4.0 at 930C cubic (masslve form)III thane 100.8 1.5 12. 4 at 9TOC platesAcetonitrile 81.6 Miacet 11.3 33.14 at 93°C plAtee

Che. Co.1.±;ttropropane 126.5 EI Pract 1.4 lO.6 at 930 C short needles2.1i1tropropane 120 MX Pract 2.3 11.6 at 930 C short needles2,14-Pentanedione 140.5 Carbide & 2.9 18.3 at 93rC flat prisms

CarbonNotbylisobutylketone 115.8 2.4 9.6 at 93 0C long prlsmnI.Propylacetate 101.6 mr Red label 1.5 6.0 at 93°C long prl=s, scme

cubica-Dutylformte 105.6 AC fed Label 1.-1 4.6 at 930C long primaEtbhvl acetate 77.1 Baker's CP 2.0 6.1 at boil. hexagonal platesa-Prapylpropionate 121 Pied Label 0.8 1.6 at 930C short prisms, some

cubicDutylacetate 126.5 2K Technical 1.1 4.0 at 930 C long prismsMethylet•ylketone 79.6 5.6 13.9 at boil. coarse plates1tiroethbane 114.2 11 Red label 3.6 19.5 at 9 30c platesIsraropylacetate 88-90" CP 1.1 3.2 at boil. long primsMesitylonide 1,8 I fled label 4.8 14.5 at 930 C platesa-Amlacetate 1146 Cp 1.0 2.1 at 930 C prismsDlmethylcarbonate 88-91 11 Red Label 1.4 6.6 at boil. platesDMetbylcarbovate 125-i26.5 E Red label 0.7 3.2 at 930 C prismIsoaawlacetate 132 CP 1.2 3.6 at 930 C primsItbylpropiomate 98-100 I Red label 3.0 10.7 at 930 C fairly thick bex

platesMethyl-a-butyrate 107 .5-103.5 EK Red label Y .,ý 4.9 at 93'JC needlesCYclopentanone 130.6 IX Red Label 1i.5 39.') at 93.5 0C hexagonal platesAcrylonitrile 7T.3 Cyanamid Co. 4.0 16.4 at boil. flat platesNmthylcellosolveacetate 144.5 Carbide & 1.6 8.8 at 93°C masive hexagona and

Carbon prisms

\

UK, Ea•tn Kodak; Pract, practical.

40

min us =:• •P'•''+ +•••'• •'+,+:s's'++++• •t• •••'• • •+ •• :+ •" n A -•' +••'- +: "

crystal seen in Figure 22b which in recrystallized from acetone by carbondioxide injection. The table shows that cubic crystals are formed fromcyclohexanone solution, and interestingly, the crystals formed by carbondioxide injection into cyclohexanone solution are also reasonably cubic as isseen in Figures 20 and 22a. As further interesting information Table IV showsthat "normal" recrystallization of RDX from two closely related solvents,viz., cyclohexanone and cyclopentanone, produces quite different crystals.

Concerning the size of crystal one might expect that bcause more RDX candissolve in cyclohexanone, at ro'm temperature, for examp!a, larger RDXparticles would result since there would be relatively more material availablefor growth after some initial nucleation. Apparently, with RDX-cyclohexanonesolutions more nuclei form at the threshold pressure relative to the eventsusing acetone under similar conditions. Figures 24a and 24b and Figures 25aand 25b are photomicrographs of RDX recrystallized from cyclohexanone withrapid injection of CO2 over a wide range of temperature and pressure. As wasshown earlier in Figure 20, very small par*icles with extremely uniform PSDcan be produced (and reproduced). The initial RDX-cyclohexanone concentrationfor this series of tests is at the saturation for the correspondingtemperatires. These photcaicrographs show that at any temperature increasingmaximum operating pressure doesn't change final particle size or PSD as longas THP is surpassed.

On the other end of the particle aize spectrum, recrystallization of RDXfrom acetone with slow injection of carbon dioxide produces very largeczrstals, and Figures 26a-and 26b show examples of some of the largest RDXcrystals produced on the program. Although at first glance the operatingconditions given appear very different, both tests' were conducted withsolutions near-saturation (at each corresponding temperature) and the maximumoper-ting pressure is the THP with a long hold time. Crystals of RDX formedfrom cyclohexanone have (except for a few particles, almost no opaque spots inthem, indicating that few intragranular cavities are formed by the gas anti-solvent recrystallization process. Similarly only very few of the RDXcrystals formed from acetone solution exhibit intragranular cavities and, ingeneral, they are very small (although in a few crystals they are quitenumerous).

As was explained earlier continuing a slow gas introduction beyond THP(for each RDX concentration and temperature) gives a product with a broadrange of particle sizes. Figures 27a, 27b. and 27c depict results ofdifferent parameter combinations, again from acetone. Particles range frommany of five micron size (barely visible in the figures) to 100-150 microns.

In some of the tests that were carried out some unusual morphology orquite peculiarly-shaped crystals were formed. Some of the shapes could beexplained and some were not, but for completeness and interest value, the nextfew figures present some quite diverse particle shapes.

On some occasions large, thin platelets were formed, and initially thereseemed to be no correlable factors to explain the platelets. Figures 28a and28b are photomicrographs of platelets recrystallized from cyclohexanone.Table IV and all the other figures which reproduced cyclohexanone-

41

FZGURI 24a - RDX CR•S'TLS - RAMID DUIECTON VIQ)GO 24b - RX CRYSTALS - RAPID INJE'ION

lIMT RD%- CYCL0HKXAM~NE INT RDI-CYCLOHEIAIIONE

(11.32 RDI. 250 C. 1000 psi) (15.01 RDX. 50 0 c. 1500 psi)

r~~ý 0 1YScc,,/

<zt (-\*2• "A, '.4

-.

2

.

,- v, " . "• . t "- -, , " "'; : "'

S,,

,-

S... "- " "

-"'"'" - "" " " - " - .• ... ",' - """ . .. :' -

-. ....

, •,..

TIGMR 25a - RDX ClYSMAS - RAPID IN3KTION FIG=R 25b - DX CIISTALS -RAPID INJECTION

INTO RDX-cT=LOHSXANON DMT RDZ-CYCLOHEA0NE

(13.5% RDZ. 50%c. 2300 psi) (13.21 RDX. 600 c. 2300 psi)

Qý 10

x%.~

CC-

7g -' 17 t

-4S, IIlk-.

47'1'44'

h

= -~ J -. o-

A-3uo* /t,

43C~

FIG= 26a - RDI CRYSTALS - I• J II3ECTION FIGURE 26b - RDx CR•YMSTS - SLO IIIECTION

IN M RDI-ACE"O IN1O RDX-ACETO NE

(12Z RWX. 350c. 1000 psi) (71 RDIX. 20 0 c. 500 pas)

.•", --•,

( I •" .. ,• :

N

A.•, , - -,2 ,- ..-

I,,.. -.

, _

171

Alt-

44

•

7"•

'-- • -• •-

• . _ .1 ~ 7•"& " .. ,4'i.

"/-,.- - \.- ¢ - •• " • f$"

7 -,e

Z7 0

," " - • . - ",,..-- 1" , -..,

Su,-",. ~~--" -%..", '•. '

I-.

Ticu.:• •,•• : . 27a Dx CRSTAS FOME MO ACTN' 50TIG0. - SLOW INECTION BEYOND

.THRESHOLD PRESSURE

(95 RX. 22,.90,piIASOLTO LWIJCIN HR

*• J•.• • .~..-. - .FIGRE.., -.•l I.oZJA- s~• • o.•R

HOL A THRESHOLD PRESSURE ,TE

•,,, .• .•.[•'7"• '" 'CONTINUTED INJEC'TION

IGE• , 2 (15 (9.61 DX. 40°2 . 1200 psi)

"f.' SOLTION. LONG HOLD TIME AT"",�". HOLDmTB ATTESOLD PRESSURE THENCO

ST INJECTION (6.8 RDX. 22C 760 ps

& 45

$ >~k~~FIGUE 2c - DX RYSTLS ORME FRM ACTON

GURI 28a - THIN PLATKRTS OF RDX OCCASIONALLY FIGURE 28b - ul PLATELETS OF RDX OCCASIONAL

FORMED n0M RDX-4CYCLOHHXA2ONE FORME FROM RDX-CYCLOEXAWON

Z,

.h;C

n;.: ... *'Ail

-.

- ~ / CC.

4.i6

I I I Ii I I

ccI

.. . . . . .- . . . . . . .. . . .. .

recrystallized RDX showed that regular (cubic) crystals are produced fromcyclohexanone solution, and thus the platelets are a priori anomalous.Crystallization tests that were carried out in the sight glass providedinformation on the (occasioLal) appearance of the platelets, and it is mostfacile to refer to a series of diagrams in Figure 29 in order to explain theformation of platelets. Figure 29a points out the initial level of PDX-cyclohexanone. and Figure 29b the level after introduction of carbon dioxideto a point short of onset, i.e., no haziness has appeared in the expandedsolution. If the gas has been introduced very rapidly, say of the order of afew seconds. the "bubbles" of carbon dioxide mixing with the cyclohexanone arelarge; the small surface area per unit weight of the gas impedes mass transferto the solution, end thus, the cyclohexanone and gas are not at equilibrium atthe top of the solution (nor in the bulk of solution). If the carbon dioxideflow is stopped and the system allowed to remain quiescent at the conditionsof Figure 29b another "zone" appears at the surface of the expanded solution.This zone is distinguished by a "refractive index interface" and is a resultof additional absorption of carbon dioxide int.o the not-completely-expandedcyclohexanone-RDX solution. The higher concentration of carbon dioxide inthis layer is responsible for the difference in refractive index which the eyecan distinguish; Figure 29c shows this zone. As the absorption and diffusioncontinues, the zone increases in thickness as depicted in Figure 29d. At"some" condition of absorption, of gas, platelets of RDX were seen to form atthe interface; when the platelets grew to a sufficient size, they disengagedfrom the diffusion layer and fell to the bottom of the sight glass cavity;Figure 29e depicts this situation, viz., the growth of the platelets, theirdisengagement from the interface, and the settling to the bottom of the sightglass.

A very peculiar collection of platelets is shown in Figure 30. Theinterior of the platelets are essentially cavity-free, but they consist of aperimeter zone of tiny cavities. (Interestingly, thin platelets in Figure 28also have tiny intragrannilar cavities.) Replication of this peculiar crystalwas not explored, but in the particular test that produced the platelets shownin Figure 30. the pressure was increased then decreased several times, i.e., apressure near the threshold pressure was reached and maintained for severalminutes, then the pressure was decreased, and the procedure repeated severaltimes. It is conjectured that platelets formed during the first pressureincrease and that during the second (or third) cycle more RDX was formedheterogeneously on the first crystals. Why the "new" RDX is so cavity-ladenis not explainable; nevertheless, the crystals are quite interesting inappearance. Other peculiarly-shaped crystals are shown in Figures 31a and31b. Figure 31a is a photomicrograph of RDX needles recrystallized fromcyclohexanone, and Figure 31b shows some needles and other crystalsrecrystallized from acetone; note the large (500 micron X 100 micron) thinplatelet in the upper right hand corner. It is barely visible in the 1.500immersion fluid, but it is seen to be absent of cavities.

In summary of all the results, the use of supercritical fluid anti-solvents has been shown to be an effective concept to recrystallize RDX fromsolution. The RDX that was used in the tests contained large cavities (cf.Figures 19a and 19b but particles containing virtually no intragranularcavities can be formed by the process, and the process can be tailored to

47

* 4

IFIGURE 29

Rfthm~tia DI ngiu of SWigt Mr-5m Testa

a b

,.High pressure(e g., 700 psig)CO2

1 al.m C

-Level of solutiono after rapid

introduction of CO202

0- Level of RDX-0cyclohexanone Bubbles of CO2solution at start 0*of test0

0

Point of introduction of CO 2

48

FIGURE 29

Schmteotl Di•w. of SMfht G1ss Tests

c d

CO 2 CO 2

In*- efrfactie afterx ~ Refractive Index1nefc after Interface after2 min.

"Expanded RDX-Expanded RDX- CyclohexanoneCyclohezanone SolutionSolution

49w

FIGMUR 29

Saefnl 1 gf 7porTMUC~n orPlatelets or LT~

e

--- Platelet forming atRefractive Index Interface

Platelets falling through-- expanded solution

panded solution

5')

FIGURE 30 - PLATELETS OF RDX FORMED FOMN

RDX-CYCLOBEXANON SOLUTION

~, /

it

51

Il N

FIGURE 31a FIGURE 31b

UNUSUAL RDX CRYSTALS FORMEDNEEDLPS OF RDX FORMED FROM RDX-

CY(fLOH NONE SOLUTION - SLOW INJECTION FROX ACETONE SOLUTIONS(61 RDX. 20°C, 800 psi) .To THESOLD PRESSURE AND HOLD (VERY(6 D.20.80pi

HIGH TEMPERATURE SOLUTION, 80°C.

201 RDX. 2500 psi)

Y , /

Ilk~

W11 v.1.ý"L1

"4., , .

/7 ~ . -. ,* . ,,• .... I.

p -

/

--

........................................... ... v"c • > , ~'"-" • "

produce RDX of controlled size and particle size distribution. Some quiteunusual crystals formed and shown in Figures 28-31 were presented primarilyfor their interest value and completeness of results obtained on the program.

There are other variations of the gas anti-solvent process that aredescribed here. Figure 21 showed that the onset of nucleation is stronglyinfluenced by the concentration level of RDX in solution. Extrapolating fromthis figure then, if two materials, say. RDX and a polymer, are dissolved in asolvent, it should be possible to co-precipitate the materials, or to coat onewith the other. For *'.e coating variation, RDX would be dissol.ved to a "high"concentration level and the polymer to a "low" concentration level. Gas canbe injected to a sufficient expansion level to recrystallize the RDX, then gasinjection continued to an expansion level sufficient to "recrystallize" thepolymer. The respective initial concentration levels are not now known, butsome preliminary feasibility tests could establish the limits. It isspeculated that the presence of the recrystallized RDX in suspension willserve as heterogenous nucleating sites for the polymer resulting in a quiteuniform coating.

The economics of the gas anti-solvent recrystallization process aredeveloped in the next section.

53

77, ... I• -- Mr. 4 777

IV. EXTmmSION OF SUPERCRITICAL FLUID hNT4I-SOLVMU RECYSTALLIZATION TOPRODUCTION SCALE OPERATION

A simplified process flow diagram for the recrystallization of RDX viathe gas anti-solvent process is given in Figure 32. In explanation of itsoperation (and with reference to the accented flow path in the figure) RDX ischarged to a dissolver (and it can be charged by means of a screw conveyor,for example), and a solvent (cyclohexanone. for example) is admitted in theappropriate ratio to dissolve the RDX. The pump, P-I, takes the RDX-cyclohexanone solution from ambient pressure to process pressure. Gas atprocess pressure is injected into the RDX-cyclohexanone solution via pump, P-4. the ratio of gas to solution set by the rates of the respective pumps. Atthe start of the sequence Valves 1 and 2 are open, 3 and 4 closed. Theexpanded solution is conveyed to the crystallizer, vessel 1. which is designedto provide sufficient residence time to produce 100 micron RDX. The filterelement serves to retain the RDX and allows the expanded liquid (cyclohexanoneand gas) to pass through. The solution passes through V-2, and is expandedslightly through the pressure reduction valve, PRV. where the solutionseparates in organic liquid and gas. The gas is liquefied and pumped by P-2to an in-process storage vessel. Solvent is recycled to the dissolver.

When the designed-for amount of RDX has been collected. Valves 3 and 4are opened. Valves 1 and 2 closed, the expanded solution leaving pump, P-1, isconveyed to Crystallizer 2. and the process of recrystallization continued asjust described. While recrystallization is continuing in Vessel 2, the RDXcrystals in Vessel 1 are washed off their adhering solvent with freshliquefied gas via pump, P-5o through valve, V-5.

Many "details" of operation and design are being ignored for thispreliminary flow sheet and process description, but the "details". of course,influence econanics, ease of operation, and the like. One method of removingthe crystals from the vessel, for example, might be via a "basket" insert inthe crystallizer. The basket might be a very thin walled cylinder fitted witha sintered metal element at the base, the entire basket placed into the vesselso as to prevent bypassing of expanded solution and RDX particles. A quickopening autoclavo top on the vessel would provide access to the interior forinserting an empty basket and removing a filled one.

irrespective of the design details, in any event, at the end of therespective recrystallization and washing segments the flow of expandedsolution is again interchanged and the process continued as previously. Theadvantage of the dual recrystallizer system is that the down-time isminimized, for example, if crystallizer 2 was not included in the design, thecrystal washing step. unloading, and insertion of a new basket would becarried out during down-time, i.e.. no production occurring.

It is felt that the two crystallizer vessel system is a reasonable onefor this preliminary cost evaluation; the methodology for developing thecapital and operating costs follows.

54

FIGU 32 - SUPIRCRICAL FLUID ANTI-SOLV'IT PRECIPITAT=ON

RDX FeedSolventMake-up

I p (P-1)

Dissolver Valve (V-1)V-i) -i V-3

P-4ISolvent (MP)RecycleStream P-5 - Crystallizer

SVV_2 Filter Elements V-4

Gas LiquefiedMake-upý Gas

Storage

"PRV

Gas-SolventSeparator

P.-3

55

a. Production level: 10.000.000 lbs/yr

b. Solubility level of RDX in cyclohexanone: 10%

c. Anti-solvent: carbon dioxide

d. Fcessure level: 800 psig max

e. Cycle time: 1 hr(Cycle time influences vessel size and cost)

f. Work schedule: 3 shifts/day. 330 days/yr

The 10.000.000 lb - 330 day production schedule calculates to 30.303lbs/day which on a Chemical Process Industries standard is not a huge volumeof material. Production figures are relatively lower, and the 30.000 lb/dayproduction rate is more related to one for fine chemicals. which RDX can beconside-ed to be. During the 1 hr cycle time about 1200 lbs of RDX arerecrysta!.'ized and collected in the basket; a 1200 lb load plus, say. 200 or300 lbs for the basket is a tractable load for a simple block and tackle orlight weight overhead crane system to handle; assuming a bulk densily of about0.8 to 1 g/cc the minimum volume of the basket would be about 20 ft ; for amore conservative design it in doublQd to 4C ft3. For a comparison, thevessels that are usid for coffee decaffeination (at the Hag plant in Bremen.FRG) are 1000(0) ft3 in volume and hold 40.000 lbe of coffee beans per fill.Furthermore. those vessels operate at 5000 psi whereas RDX recrystallizationoperates at only 150 psi. As further comparison of processes the volume oforganic solution that is pumped from the dissolver (by Pump. P-1) is only 25gpa; a "few" horsepower motor is more than sufficient to provide a 150 psihead. For the coffee plant the solvent recirculation pump is 5000 Hp (and itis for the ability to provide these comparisons that the background andapplications of supercritical fluids was given in some detail in Section II).

Phasex Corporation formed an association with the Glitsch Package PlantsDivision in 1986; during the last year and one-half three skid mountedsupercritical fluid plants have been designed and construction of one of themis almost completed. The cost for small plants, of the order of a few millionlbs/yr, is. therefore. reasonably well known. Figure 33 is a very detailedpiping and instrumentation diagram (P/ID) for one of the plants. The "nutsand bolts" design or even the detail required for the P/ID was outside thescope of the Phase I progra; however, based upon the similarities between theRDX recrystallization process and the other processes the cost of the10.000.000 lbs/yr recrystallization plant is estimated at about $5.OMM.

In order to arrive at a recrystallization cost. then. bases a-f. a plantcost of $5.0M4. and the following additional conditions are stipulated:

56

/(

31

0- I

3j

lotI

Isma~mmoo

1. 4 persons/shift = 3.960 person days/yr

2. Salary plus fringes assumed to be $150/person day;Labor costs = 3.960 x 150/10M = 36€/lb

3. Depreciation schedule. 5 yr straight line;Capital costs = $ONM/(IOMM lbe z 5 yrs) = I0C/lb

4. Buildings. facilities. 20% of capital costs : 20/1b

5. Maintenance. 1OZ of capital costs = 1.2C/lb

6. Insurance. taxes. 10Z of capital costs = 1.2€/lb

7. Supervision. materials flow accounting for transfer costs. 100% oflabor costs = 6€/lb

8. Gas and solvent make-up. utilities; no &as and solvent make-up valueswere evaluated on Phase I but for conservatism a high cost will beestimated, viz.. 20€/lb

Trble III summarizes the costs.

Table III

Capital and Operating Costs for Gas Anti-Solvent Recrystallization of RDX(10.000.000 lbs/yr)

Capital Costs

Plant and buildings 12€/lb

Operatim Costs

Labor 6

Maintenance 1.2

Insurance. Taxes 1.2

Supervision. Accounting 6

Make-up, Utilities 20

Total Costs 471/lb

Since this estimate is of a preliminary nature, it is suggested that acost rangn of about $0.50 to $1.00/lb RDX be considered as the capital andoperating cost for recrystallizing RDX.

58

- C .

V. CONCUSIONS AND U=)IZDAfONS

A. Conclusios

1. Recrystallization of RDX by Supercritical Fluid Anti-Solvents hasbeen demonstrated to be an effective technique to form RDX free ofintragranular cavities.

2. With appropriate combinations of parameters particle size can becontrolled; RDX crystals of quite narrow particle size distributioncould be formed ranging in average particle size from about 10microns to about 100 microns. The 10 micron smie RDX is verydifficult to produce by the conventional Holaten AAP process.

3. A preliminary economic analysis shoved that at a processing level of10.000.000 lbs/yr of RDX. the supercritical fluid anti-solventprocess could recrystallize RDX at total capital and operating costsof about $0.50 to $1.00/lb.

4. Other capabilities of the supercritical fluid anti-solvent processwere suggested, viz.. co-precipitation of explosives and othermaterials and the coating of explosive crystals with polymers.

B. Recommendations

1. Because of the positive results obtained on the Phase I program, aPhase II study should be carried out.

2. The other processing potentials of supercritical fluid anti-solventsdescribed should be evaluated at the feasibility level.

59

VI. LITERATURE CITATIONS

1. Booth. H.S. and R.M. Bidwell. Solubility Measurements in the CriticalRegion. Chem. Rey.. 44. 477-513 (1949).

2. Paul. P.M.F. and W S. Wise. The Principles of Gas Extraction. Mills andBoon LTD. London. 1971.

3. Irani. C.A. and E.W. Funk, Separation Using Supercritical Gases. in CRCHandbook: Recent Developments in Separation Science. Vol. III. Part A.CRC Press, Boca Raton. FL. 171-179 (1977).

4. Paulaitis, M.E., V.J. Krukonis, R.T. Kurnik. and R.C. Reid, SupercriticalFluid Extraction, Rev, in Chem. Eng., 1(2). 179-250 (1983).

5. Diepen. G.A.M. and F.E.C. Scheffer. On Critical Phenomena of SaturatedSolutions in Binary Systems. J. Am. Chem. Soc.. 4081-5 (1948).

6. Diepen. G.A.M. and F.E.C. Scheffer. The Solubility of Naphthalene inSupercritical Ethylene. J. Am. Chem. Soc.. 70. 4085-9 (1948).

7. Diepen, G.A.M. and F.E.C. Scheffer. The Solubility of Naphthalene inSupercritical Ethylene II. J. Phys, Chem.. 57. 575-8 (1953).

8. MacKay. M.E. and M.E. Paulaitis, Solid Solubilities of Heavy Hydrocarbonsin Supercritical Solvents. Ind. Eng. Chem. Fund.. 18. 144-53 (1979).

9. McHugh. M.A. and M.E. Paulaitis. Solid Solubilities of Naphthalene andBiphenyl in Supercritical Carbon Dioxide. J. Chem. Eng. Data, 25. 326-9(1980).

10. Kurnik. R.T. and R.C. Reid. Solubility of Solid Mixtures in SupercriticalFluids. Fluid Phase Equilibria, 8. 93-105 (1982).

11. Hannay, J.B. and J. Hogarth. On the Solubility of Solids in Gases, Proc.Roy. Soc.. (London), 29. 324-26 (1879).

12. Tsekhanskaya. Yu.V., M.B. Iomtev, and E.V. Mushkina. Solubility ofDiphenylamine and Naphthalene in Ethylene and Carbon Dioxide UnderPressure. Russ., J. Phys. Chem.. 36. 1177-81 (1962).

13. Tsekhanskaya. Yu.V., M.B. Iomtev. and E.V. Mushkina. Solubility ofNaphthalene in Ethylene and Carbon Dioxide Under Pressure. Russ. J. Phys.Chm.. 38. 1173-6 (1964).

14. Tsekhanskaya. Yu.V., Diffusion of Naphthalene in Carbon Dioxid NearLiquid-Gas Critical Point. Russ. J. Phys. Chem.. 45. 744-44 (1971).

15. King, A.D.. and W.W. Robertson. Solubility of Naphthalene in CbmpressedGases. J. Chem. Phys.. 37, 1453-5 (1962).

60

I IIIIII I••11•,,K

16. Najour. G.C. and A.D. King, Jr.. Solubility of Naphthalene in CompressedMethane. Ethylene and Carbon Dioxide., J. Chem. Phys.. 45. 1915-21 (1966).

17. Schmitt, W.J. and R.C. Raid. Solubilities of Several Model OrganicCompounds in Chemically Different Supercritical Fluids. Annual Meeting,AIChE. San Francisco. Ncvember 1984.

18. Krukonis, V.J.. M.A. McHugh. A.J. Seckner. Xenon as a SupercriticalSolvent. J. Phys. Chem.. 88. 2687-9 (1984).

19. McHugh. M.A.. A.J. Seckner. and V.J. Krukonis, Supercritical Xenon,Annual Meeting. AIChE. San Francisco. November 1984.

20. McHugh. M.A.. A.J. Seckner. and T.J. Yogan. High Pressure Phase Behaviorof Mixtures of Carbon Dioxide and Octacosane. IEC Fundamentals. 23, 493-496 (1984).

21. Najour. G.C. and A.D. King. Jr.. Solubility of Anthracene in CompressedMethane, Ethylene, and Carbon Dioxide: The Correlation of Anthracene-gas Second Virial Coefficients Using Pseudocritical Parameters. J. Chem.Phys.. 52. 5206-11 (1970).

22. Eisenbeiss. J.. A Basic Study of the Solubility of Solids in Gases atHigh Pressures. Final Report. Contract No. DA18-108-AMC-244(A). SouthwestResearch Inst., San Antonio. (August 6. 1964).

23. Kurnik. R.T.. S.J. Holla. and R.C. Reid. Solubility of Solids inSupercritical Carbon Dioxide and Ethylene. J. Chem. Eng. Data, 26. 47-51(1981).

24. Kennedy. G.C.. A Portion of the System Silica-Water. Econ, Geol., 45.629-36 (1950).

25. VanLeer. R.A. and M.E. Paulaitis. Solubilities of Phenol and ChlorinatedPhenols in Supercritical Carbon Dioxide. J. Chem. Eng. Data. 25. 257-9(1980).

26. Holden, C.H. and 0. Maass. Solubility Measurement in the Region of theCritical Point. Can. J. Res. 18B. 293-304 (1940).

27. Ewald. A.H.. The Solubility of Solids in Gases. Trans. Far. Socl. 49.1401-85 (1953).

29. Todd. D.B.. and J.C. Elgin. Phase Equilibria in Systems with EthyleneAbove -:ts Critical Temperature. AIChE J.. 1. 20-7 (1955).

30. Caragay. A.B.. Supercritical Fluids for Extraction of Flavors andFragrances from Natural Products. Perf. and Flavors. 6. 43-55 (1981).

31. Schultz. W.G.. J.N. Randal. Liquid Carbon Dioxide for Selective AromaExtraction. Food Tech.. 24. 94-8 (1970).

61

32. Randall. J.M., W.G. Schultz. A.I. Morgan, Jr., Extraction of Fruit Juices.and Concentrated Essences with Liquid Carbon Dioxide. Confructa, 16. 10-16 (1971).

33. Anon.. A Supercritical Fluid Extraction Alternative to Distillation ofAlcohol, Chem. Eng.. 88(9). 19 (1981).

34. Paulaitis, M.E.. M.L. Gilbert, and C.A. Nash. Separation of Ethanol-Water Mixtures with Supercritical Fluids, 2nd World Congress of ChemicalEngineering. Montreal, October 1981.

35. Moses. J.M.. K.E. Goklen, and R.P. deFilippi, Pilot Plant Critical-FluidExtraction of Organics from Water, Annual Meeting. AIChE. Los Angeles.November 1982.

36. McHugh. M.A.. M.W. Mallet, and J.P. Kohn, High Pressure Phase Equilibriaof Alcohol-Water-Supercritical Solvent Mixtures. Annual Meeting. AIChE,.New Orleans, November 1981.

37. Modell, M.. R.P. deFilippi., and V.J. Krukonis. Regeneration of ActivatedCarbon with Supercritical Carbon Dioxide, Amer. Chem. Soc. Meeting.Miami. September 1978.

38. Modell, M.. R.J. Robey. V.J. Krukonis, and R.P. deFilippi. SupercriticalFluid Regeneration of Activated Carbon. 87th Nacional Meeting, AIChE.Boston, August 1979.

39. Zosel. K., Separation with Supercritical Gases: Practical Applications,Angew. Chem. Int. Ed, Engl.. 17. 702-15 (1978). ., •.

40. Roseluis, W.. 0. Vitzthum, and P. Hubert. Method for the Production of.Caffeine-Free Coffee Extract, U.S. Patent 3.843,824; 22 October 1974.

41. Prasad. R.. M. Gottesman, and R.A. Scarella. Decaffeination of AqueousExtracts, U.S. Patent 4,246,291; 20 January 1981.

42. Margolis. G. and J. Chiovini. Decaffeination Process. U.S. Patent4,251.559; 17 February 1981.

43. Friedrich, J.P.. G.R. List, and A.J. Leakin. Petroleum - Free Extractionof Oil from Soybeans with Supercritical CO2V J. Amer. Oil Chem. Soc..59(7), 288-92 (1982).

44. Christianson, D.D., J.P. Friedrich, G.R. List, K. Warner, E.B. Bagley.A.C. Stringfellow. and G.E. Inglett. Supercritical Fluid Extraction ofDry-Milled Corn Germ with Carbon Dioxide, J. Food Sci.. 49(1) 229-32 and272 (1984).

45. Stahl, E., E. Shutz. and H.K. Mangold. Extraction of Seed Oils withLiquid and Supercritical Carbon Dioxide, Ag. Food Chem., 28. 1153-7(1980).

62

46. Anon.. Extraction Process Benefits Potato Chips. Coffee. Hops. Food Eng.Int.. 45-6. uctober 1981.

47. Wise. W.S.. Solvent Extraction of Coal. Chem. Ind.. (Lon.,n). 950-1(1970).

48. Gearhart. J.A. and L. Garwin. Resid-Extraction Process OffersFlexibility, Oil Gas J., 74(24). 63-6 (1976).

49. Gangoli. N. and G. Thodas. Liquid Fuels and Chemical Feedstocks from Coalby Supercritical Gas Extraction. IEC. Prod. Res. Dev.. 16(3). 208-16(1977).

50. Stahl, E., W. Schilz. E. Schutz. and E. Willing. A Quick Method for theMicroanalytical Evaluation of the Dissolving Power of SupercriticalGases. Angew. Chem. Int. Ed. Engl.. 17. 731-8 (1978).

51. Krukonis. V.J.. A.R. Branfman. and M.G. Broome. Supercritical FluidExtraction of Plant Materials Containing Chemotherapeutic Drugs. 87thAIChE National Meeting. Boston. August 1979.

52. Fremont. H.A.. Extraction of Coniferous Woods with Fluid Carbon Dioxideand Other Supercritical Fluids. U.S. Patent 4.308.200; 20 December 1981.

53. Anon.. Gas Solvents: About to Blast Off. Bus. Week. July 27. 68M (1961).

54. Cookson. C.B.. Subcritical Extractions Using LiqLid CO2 , Spring Tech.Symp. Bulk Pharma. Chemicals. Newport. RI. May 1987.

55. Perkins. B., Food Without Cholesterol Close to Reality. The CincinnatiEnquirer. Jan. 10. Dl (1987).

56. Wilke. G.. Studiengesellschaft Kohle. Mulkeim (Ruhr). W. Germany. privatecmnmunication.

57. Krukonis. V.J., Processing of kolymers with Supe-critical Fluids. PolymerNews. 11. 7-16 (1985).

58. Krukonis. V.J.. Supercritical Fluid Fractionation: An Alternative toMolecular Distillation. Nat. Mtg. AIChE, Houston. March 1983.

59. McHugh. M.A. and V.J. Krukonis. Supercritir.al Fluid Extraction:Principles and Practice. Ch.9. Butterworth Publishers. Boston (1986).

60a. Worthy. W.. Supercritical Fluids Offer Improved Separations, C&E News.Aug. 3. 16-17 (1981).

60b. Paulaitis. M.E.. V.J. Krukonis. R.T. Kurnik. and R.C. Reid. SupercriticalFluid Extraction. Rev. Chem. Eng. I(2). 179-250 (1983).

60c. Krukonis. V.J.. Supercritical Flul'd Nucleation of Difficult-to-ComminuteSolids. Ann. Mtg. AIChE. San Francisco. November 1984.

63

61. Larson, K.A. and M.L. King. Evaluation of Supercritical Fluid Extractionin the Pharmaceutical Industry. Biotech. Prog., 2.. 73-82 (1986).

62. Petersen. R.C.. D.W. Matson, and R.D. Smith. Precipitation of PolymericMaterials from Supercritical Fluid Solution. Polymer Prepr., 27. 261-62(1987).

63. Elgin. J.C.. and J.J. Weinstock. Phase Equilibria at Elevated Pressuresin Ternary Systems of Ethylene and Water with Organic Liquids: SaltingOut with a Cupercritical Gas, J. Chem. Eng. Data 41, 3-10 (1959).

64. Paulaitis. M.E.. J.R. DiAndreth, and R.G. Kander. An Experimental Studyof Phase Equilibria for Isopropanol-Water-0 2 Mixtures Related toSupercritical Fluid Extraction of Organic Compounds from AqueousSolgtions, in Supercritical Fluid Technology. J.M.L. Penninger. M.Radosz, M.A. McHugh. and V.J. Krukonis, ads.. Elsevier SciencePublishers, Amsterdam (1985).

65. Irani, C.A.. C. Cosewith, and S.S. Kasegrande, New Methods for HighTemperature Phase Separation of Solutions Containing CopolymerElastomers. U.S. Patent 4,319.021 (1977).

66. McHugh. M.A. and T.L. Gucken. Separat.'ng Polymer Solutions withSupercritical Fluids, Macromolecules 18. 674-81 (1985).

67. Gibbs. J.W., The Collected Works of J.W. Gibbs. Vol. I TLermodynamicr. p.

322. Yale University Precs, New Haven,. CT (1957).

68. Adamson, A.W.. Physical Chemistry of Surfaces, pp. 288-2M8. IntersciencePublishers, Easton. PA (1963).

64>

DISTRIBUTION LIST

No of No ofQ Q mn Qrgani•adn

12 AdministratorDefense Technical Info CenterATIN: DTIC-DDACameron StationAlexandria, VA 22304-6145

HQDA (SARD-TR)Washington, DC 20310-0001

CommanderUS Army Materiel CommandATTN: AMCDRA-ST 1 Commander5001 Eisenhower Avenue US Army Missile CommandAlexandria, VA 22333-0001 ATTN.: AMSMI-AS

Redstone Arsenal, AL 35898-5010I Commander

US Army Laboratory Command 1 CommanderATIN: AMSLC-DL US Army Tank Automotive CommandAdelphi, MD 20783-1145 A'TTN: AMSTA-DDL (Technical Library)

Warren, MI 48397-50002 Commander

Armament RD&E Center 1 DirectorUS Army AMCCOM US Army TRADOC Analysis CommandXITN: SMCAR-MSI ATrN: ATAA-SLPicatinny Arsenal, NJ 07806-5000 White Sands Missile Range, NM 88002-5502

2 Commander 1 CommandantArmament RD&E Center US Army Infantry SchoolUS Army AMCCOM ATIN: ATSH-CD-CSO-ORATTN: SMCAR-TDC Fort Benning, GA 31905-5660Picatinny Arsenal, NJ 07806-5000

1 AFWL/SULDirector Kirtland AFB, NM 87117-5800Benet Weapons LaboratoryArmament RD&E Center I Air Force Armament LaboratoryUS Army AMCCOM ATiN: AFATL/DLODLATrN: SMCAR-LCB-TL Eglin AFB, FL 32542-5000Watervliet, NY 12189-4050

1 ChairmanCommander DOD Explosives Safety Board"US Army Armament, Mmitions Room 856-C

and Chemical Command Hoffman Bldg IATrN: SMCAR.ESP-L 2461 Eisenhower AvenueRock Island, IL 61299-5000 Alexandria, VA 22331

Commander I CommanderUS Army Aviation Systems Command US Army Armament RD&E CenterATTN: AMSAV-DACL ATrN: SMCAR-LCE, Dr. N. Slagg4300 Goodfellow Blvd. Picatinny Arsenal, NJ 07806-5000St. Louis, MO 63120-1798

I CommanderDirector US Army Armament, Munitions &US Army Aviation Research Chemical Command

and Technology Activity ATIN: AMSMC-IMP-L ,Ames Research Center Rock Island, IL 61299-7300Moffett Field, CA 94035-1099

65

,,1

ell

]\ i ,. __L..--.• ___./_.

A _- 7777

DISTRIBUTION LIST

No of No ofCA~i Qrganizadgn C~v Qrgzain~

Commander 1 CommanderBelvoir RD&E Center Naval Sea Systems CommandATTN: STRBE-JMC, Mr. Rick Marinez ATTN: Dr. R. BowenFort Belvoir, VA 22060-5606 SEA 061

Washington, DC 20362CommanderUS Army Aviation Systems Command 1 CommanderATTN: AMSAV-ES Naval Explosive Ordnance Disposal4300 Goodfellow Boulevard Technology CenterSt. Louis, MO 63120-1798 ATTN: Technical Library

Code 604Commander Indian Head, MD 20640CECOM R&D Technical LibraryATIN: AMSEL-IM-L (Reports Section) 1 Commande&

B. 2700 Naval Research LabFort Monmouth, NJ 07703-5000 ATTN: Code 6.00

Washington, DC 20375I Director

Missile and Space Intelligence Center 1 CommanderATTN: AIAMS-YDL Naval Surface Weapons CenterRedstone Arsenal, AL 35898-5500 ATrN: Code G13

Dahlgren, VA 22448-5000CommanderUS Army Missile Command 1 CommanderATrN: AMSME-RK, Dr. R.G. Rhoades Naval Surface Weapons CenterRedstone Arsenal, AL 35898 ATIN: Mr. L. Roslund, RIOC

Silver Sping, MD 20903-5000Commandant

USAOMMCS 1 CommanderATTN: ATSK-CME, S. Herman Naval Surface Weapons CenterRedstone Arsenal, AL 35897-6500 ATTN: Mr. M. Stosz, RI01

Silver Spring, MD 20903-5000CommanderUS Army Missile Command I CommanderATTN: AMSMI-RO-PR-T Naval Surface Weapons Center

Dr. Porter Mitchell ATTN: C. Gotzmer, RllRedstone Arsenal, AL 35898 Indian Head, MD 20640-5000

Commander 1 CommanderUS Army Development & Employment Naval Surface Weapons Center

Agency ATTN: Dr. R.C. Gill, Bldg 600ATIN: MODE-ORO Indian Head, MD 20640-5000Fort Lewis, WA 98433-5000

1 Commander2 Commander Naval Surface Weapons Center

US Army Research Office ATTN: Code X211, LibATTN: Chemistry Division Silver Spring, MD 20903-5000P.O. Box 12211Research Triangle Park, NC 27709-2211 1 Commander

Naval Surface Weapons Center2 Office of Naval Research ATTN: R.R. Bemecker, R13

ATIN: Dr. A. Faulstich, Code 23 Silver Spring, MD 20903-5000800 N. Quincy StreetArlington, VA 22217 1 Commander

Naval Surface Weapons CenterATIN: J.W. Forbes, R13Silver Spring, MD 20903-5000

66

WILK 71Z,

DISTRIBUTION LIST

No of No ofQf in Oraiai Quiz Q izatign

Commander 1 CommanderNaval Surface Weapons Center Air Force Rocket PropulsionATTN: S. J. Jacobs, RIO LaboratorySilver Spring, MD 20903-5000 A1TN: Mr. R. Geisler,

Code AFRPL MKPACommander Edwards AFB, CA 93523 -.Naval Surface Weapons Center ,ATrN: J. Short, R12 3 AFATLIMNESilver Spring, MD 20903-O000 ATTN: G. Parsons

M. PatrickCommander Eglin AFB. FL 32542-5000Naval Weapons CenterATTN: L H. Josephson, Code 326B 1 CommanderChina Lake, CA 93555 Ballistic Missile Defense Advanced,.-

Technology CenterCommander ATTN: Dr. David C. SaylesNaval Weapons Center P.O. Box 1500ATTN: M. Chan, Code 3891 Huntsville, AL 35807China Lake, CA 93555

I DirectorCommander Lawrence Livermore National LaboratoryNaval Weapons Center University of CalifarniaATrN: Dr. B. Lee, Code 3264 ATTN: Dr. M. FingerChina Lake, CA 93555 P.O. Box 8CG

Livermore, CA 94550CommanderNaval Weapons Center 1 DirectorATTN: Dr. L. Smith, Code 326 Lawrence Livermore National LaboratoryChina Lake, CA 93555 University of California

ATFN: Dr. R. McGuireCommander P.O. Box 808Naval Weapons Center Livermore, CA 94550ATrN: Dr. R. Atkins, Code 385China Lake, CA 93555 1 Director

Los Alamos National LaboratoryCommander ATlN: Mr. J. RamseyNaval Weapons Center P.O. Box 1663ATrN: Dr. R. Reed, Code 388 Los Alamos, NM 87545China Lake, CA 93555

1 DirectorCommander Los Alamos National LaboratoryNaval Weapons Center ATIN: Dr. L. StretzATTN: Dr. K.J. Graham, Code 3891 P.O. Box 1663China Lake, CA 93555-6001 Los Alamos, NM 87545

Commander 1 Sandia National LaboratoriesNaval Weapons Station ATTN: Dr. E. Mitchell, SupervisorNEDED Division 2513ATrN: Pr. Louis Rothstein, Code 50 P.O. Box 5800Yorktown, VA 23691 Albuquerque, NM 87185

Commander 10 Central Intelligence AgencyFleet Marine Force, Atlantic OIR/DB/StandardATFN: G-4 (NSAP) GE47 HQNorfolk, VA 23511 Washington, DC 20505

67

/ 7- 7 7

DISTRIBrIMON LIST

No ofc Qgwm~

I Irving B. AkstIDOS corporationP.O. Box 285Pampa, TX 79065

I Holston Defense CorporationATIN: Did. Mahaffey, SuperintendentKingspor IN 37660

Aberdeen Proving GroundDir, USAMSAA

A MXYTN:P H. Cohen

Cdr, USATECOMATMN: AMSTE-TO-F

AMSTE-SI-FCdr, CRDEC, AMCCOM

ATTN: SMCCR-RSP-ASMCCR-MUSMCCR-SPS-ELSMCCR.OPP. C. Hassell

68

7777ý=1,

lz ~