Propellants, Explosives, Pyrotechnics 24, 255Ä…259 (1999) 255
Quantitative Characterization of Internal Defects in RDX Crystals
Lionel Borne and Jean-Claude Patedoye
French-German Research Institute of Saint-Louis (ISL), F-68301 Saint-Louis Cedex (France)
Christian Spyckerelle
 Â
Societe Nationale des Poudres et Explosifs (SNPE), F-84706 Sorgues Cedex (France)
 Â
Quantitative Charakterisierung der Fehlstellen im Innern von Caracterisation quantitative des defauts internes dans des cristaux
Á
RDX-Kristallen d'hexogene
È Â Â Â
Groûe, Gestalt, Fehlstellen sind sehr wichtige Eigenschaften von La taille, la forme, les defauts internes sont des proprietes impor-
Á
Explosivstoffkristallen. Diese Parameter spielen eine Rolle sowohl bei tantes des cristaux d'explosif. Ces parametres inŻuencent d'une part la
der Herstellung als auch beim detonativen Verhalten von Spreng- fabrication des compositions explosives, et d'autre part le comporte-
 Ã
stoffformulierungen. Die Verwendung von Kristallen frei von ment detonique de ces memes compositions. L'emploi de cristaux
È È È Â Â
Losungsmitteleinschlussen fuhrt zur Abnahme der Schlag- d'explosif contenant peu de defauts internes permet de reduire la
 Â
emp®ndlichkeit bei gegossenen Sprengstoffformulierungen. Viele sensibilite au choc de formulations explosives coulees. Des tentatives
    Â
Anstrengungen zur Herstellung solch hochwertiger Explosiv- pour fabriquer des cristaux d'explosif sans defauts ont ete realisees et
stoffkristalle sind durchgefuhrt worden und werden noch weitergefuhrt des travaux sont en cours. La microscopie optique avec adaptation
È È
È Â
Qualitative Beobachtungen von Kristallen mit Fehlstellen konnen d'indice permet une observation qualitative de certains defauts inter-
durchgefuhrt werden mittels optischer Mikroskopie und Vergleich mit nes dans les cristaux tels que des cavites occluses. L'objectif de notre
È Â
  Â
dem Brechungsindex. Ziel dieser Arbeit ist, zwei genaue quantitative travail est de decrire deux methodes quantitatives de caracterisation de
Methoden fur Messungen der Kristallfehlstellen zu entwicheln. Die ces cavites internes dans les cristaux d'explosif. La premiere methode
È Â Á Â
Á Â Â
erste Methode basiert auf genauen Messungen der scheinbaren Kri- repose sur une mesure tres precise de la densite apparente des cristaux.
stalldichte. Die zweite Methode bestimmt die Masse der Einschlusse La seconde methode permet de mesurer la masse des especes con-
È Â Á
im Kristallinnern. Die Versuche wurden durchgefuhrt mit zwei RDX- tenues dans les cavites internes aux cristaux. Des mesures realisees sur
È Â Â Â
Á Â Â
Chargen. Die strenge Korrelation, die zwischen den Ergebnissen der deux lots de cristaux d'hexogene sont presentees. Les deux techniques
 Â
beiden Methoden auftritt, beweist die Richtigkeit der Messungen. Die de mesure des defauts internes fournissent des resultats com-
È Â Â Â
Meûmethoden sind komplementar. Die Dichtemessungen liefern eine plementaires. La forte correlation obtenue entre les resultats issus des
genaue allgemeingultige Charakteristik der Fehlstellenanhaufung einer deux methodes de mesure valide les techniques et protocoles experi-
È È Â Â
Â
Kristallcharge und erlauben eine Eingruppierung der Kristalle nach mentaux. La mesure de la densite apparente des cristaux fournit une
Â
ihrer Scheindichte. Mit der zweiten Methode werden die einge- information quantitative globale sur les populations de defauts internes
 Â
schlossenen Fremdsubstanzen bestimmt. aux cristaux. Cette methode de mesure est egalement un outil de tri des
 Â
cristaux en fonction de leur densite apparente. La seconde methode
Á
permet l'identi®cation et la quanti®cation des especes chimiques
contenues dans les cristaux.
Summary crystal quality is a possible solution. Some important crystal
properties are studied and pointed out in several published
Size, shape, internal defects are very important properties of
studies. The crystal size(1), the crystal surface and shape(2,3)
explosives crystals. These parameters play a role on both the explosive
and the internal crystal defects population(4Ä…6) are important
formulation processing and the detonic behavior of the explosive
formulations. The use of explosive crystals free of solvent inclusions crystal properties which play a role on the shock to detona-
leads to decrease the shock sensitivity of cast explosive formulations.
tion transition of cast explosive formulations. These crystal
Many efforts for processing such high quality explosive crystals have
parameters are also important for formulation processing.
been done and are still in progress. Qualitative observations of internal
They can modify the bonding between the crystals and the
crystal defects can be performed by optical microscopy with matching
refractive index. The purpose of this paper is to provide two accurate
surrounding (the polymer for PBX's). Many efforts have
quantitative tools for internal crystal defects measurements. The ®rst
been done and are in progress to ®nd methods of processing
method is based on accurate measurements of the crystal apparent
explosive crystals with the suited properties(7Ä…10).
density. The second method records the mass of the species entrapped
All these works need some metrology tools to quantify the
in the crystal internal cavities. Experiments are performed on two
RDX batches. The strong correlation recorded between the results of
explosive crystal properties. Measurements of the crystal size
the two complementary methods validates the measurements. Appar-
distributions are performed using several tools: sieving,
ent density measurements provide an accurate global characterization
Coulter counter, laser diffraction or image analysis. Gas
of the internal defects population of a crystal batch sorting the crystals
adsorption methods and mercury intrusion porosimetry are
in function of their apparent density. The second method is a tool to
identify the species entrapped in the crystals.
quantitative experiments to characterize crystal shape and
crystal surface. The aim of this paper is to present quantita-
tive tools to characterize the internal crystal defects popula-
1. Introduction
tions. An accurate record of the crystal apparent density is
proposed to measure the global amount of the internal
To reduce the vulnerability of explosive formulations,
defects(4,6). Fine measurements of the mass of the entrapped
several parameters can be tuned. Improving the explosive
# WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999 0721-3115/99/0306Ä…0255 $17.50‡:50=0
256 L. Borne, J. C. Patedoye, and C. Spyckerelle Propellants, Explosives, Pyrotechnics 24, 255Ä…259 (1999)
species in the internal cavities are performed. Based on two particle. This gives the accuracy required to be able to
dedicated RDX batches, the two kinds of measurements give characterize the global amount of internal defects in a
strongly correlated results, underlining the accuracy of the crystal using the record of the crystal apparent density. An
experiments. RDX crystal internal cavities are mainly ®lled accuracy of 0.0001 g=cm3 for the apparent density measure-
with air, water and recrystallization solvent. ments will allow the detection of an internal defect of 10 mm
®lled with air in a particle with a diameter of 250 mm.
The crystal apparent density measurement is performed
using a Żotation method. Crystals are immersed into a
2. RDX Batches
mixture of toluene and methylene iodide (CH2I2) to sort the
crystals of higher apparent density in the lower part of the
Two dedicated RDX batches have been used. The manu-
container. The mixture density can be tuned between
facturing processes of the two batches are similar but two
0.9 g=cm3 and 3.3 g=cm3. The main points of the experiment
different solvents are used: acetone and cyclohexanone.
are:
Particle sizes of both batches are located between 200 mm
and 300 mm. In both cases the water used during the crystal
A good accuracy for the mixture density measurements.
processing has been marked by adding potassium bromide
This is achieved using a PAAR densimeter using the
(KBr). This allows to measure the amount of water entrapped
vibrating tube principle. The accuracy is 0.0001 g=cm3.
in the crystal internal cavities.
A good homogeneity and temperature stability of the
Figure 1a and Figure 1b are obtained using optical
Żotation mixture. A double wall container and a Żow of
microscopy with matching refractive index. These qualita-
water between the two walls limit the residual tempera-
tive observations show the internal defects populations of
ture variations below 0.1 C.
different crystals. These crystal internal cavities are the
The crystal solubility in the liquid mixture must be lim-
macroscopic part (optically observable) of larger crystal
ited. The Żotation mixture must exhibit a good wetting
defects(11). Baillou and co-workers have shown that the
for the crystals.
internal cavities contain recrystallization solvent and
The experiment starts using a liquid density higher than the
mineral salt(5).
apparent density of any particle.
Then the following steps are performed until all the
crystals have been extracted from the container:
3. Experimental Tools
Particles are dispersed in the mixture by stirring;
3.1 ISL Flotation Method and Apparent Density separation of crystals is performed by decanting. The
Measurement decanting time is a function of the crystal size. Two hours
is a convenient time for 200 mm crystals;
Table 1 gives the expected apparent density variation the lower part of the container is extracted, washed with
generated by a spherical cavity in an academic spherical toluene, dried and weighted;
Figure 1a. RDX 200=300 mm Ä… recrystallization with acetone. Figure 1b. RDX 200=300 mm Ä… recrystallization with cyclohexanone.
Propellants, Explosives, Pyrotechnics 24, 255Ä…259 (1999) Quantitative Characterization of Internal Defects in RDX Crystals 257
Table 1. Apparent Density Variations for a Spherical Particle of Diameter D with a Spherical Cavity of Diameter d
(Material density: 1.8 (RDX), air density: 0.001205, acetone density: 0.792)
Dd Dr Dr Dr
(mm) (mm) (cavity ®lled with air) (cavity ®lled with water) (cavity ®lled with acetone)
(g=cm3)(g=cm3)(g=cm3)
250 5 1.4E-05 0.6E-05 0.8E-05
250 10 11.4E-05 5.1E-05 6.5E-05
200 40 143E-04 64E-04 80E-04
250 40 73E-04 32E-04 41E-04
300 40 42E-04 19E-04 23E-04
250 80 585E-04 262E-04 330E-04
A small volume of toluene is added to the decanted First it is important to identify the species entrapped in the
Żotation mixture in order to decrease its density of crystals. It is reasonable to assume that in internal crystal
0.001 g=cm3. Then the previous three steps are executed cavities we may ®nd species present at the time of the
again. formation of the crystal: recrystallization solvent and water.
We proved also that air may be entrapped. One crystal-
The experimental results are given by plotting on the Y-
lization experiment was realized under pure argon atmo-
axis the cumulative weight in percent of crystals with an
sphere; the gas released after dissolution of the crystals was
apparent density higher than the given value on the X-axis.
identi®ed as argon using gas chromatography.
Figure 2 gives the experimental results for the two RDX
The quantity of entrapped crystallization solvent is deter-
batches. The four measurements performed on each batch
mined by gas chromatography (internal standard method)
give the reproducibility of the experiment. Its accuracy will
after dissolution of the crystals in a solvent (N-methyl
be discussed later. The use of acetone as the recrystallization
pyrrolidone).
solvent in this case leads to a smaller crystal apparent density
An indirect method was set up for the determination of the
than the use of cyclohexanone.
water content of the two dedicated batches. The water used
During each experiment, crystals are sorted according to
during the crystal processing was marked by adding a de®ned
their apparent density. For the two RDX batches, 11 classes
amount of potassium bromide (KBr). For water content
of crystals with a narrow apparent density distribution have
measurement, crystals are dissolved in a solvent (cyclohex-
been obtained (Table 2). The amount of entrapped species
anone), and the solution is extracted by water. Water
would be measured for crystals of most of the classes.
entrapped in the crystals is deduced from the bromide ion
concentration measured by ion chromatography in the
extract. Results have been compared to those obtained by
3.2 SNPE Entrapped Species Measurements the standard Karl Fischer determination on some samples,
and a good correlation has been found.
Table 3 gives the porous volume for several academic A special glass apparatus was designed to measure micro-
spherical particles with spherical cavities of various sizes. liters of gas evolved. Presence of air in the crystals has been
This leads to the amount of particles needed to get a total evidenced by volumetric measurements on some samples,
porous volume of 1 ml. This illustrates the accuracy required but because of a lack of accuracy of the method for small
to set up the measurements of the entrapped species. Appro- volumes, air content was determined using volume computa-
priate methods have been developed to search and to measure tions.
the traces of the entrapped species.
Table 2. Various Classes of RDX
Crystals Processed for Each RDX Lot
Classes
1:80005r g=cm3
1:79905r51:8000 g=cm3
1:79805r51:7990 g=cm3
1:79705r51:7980 g=cm3
1:79605r51:7970 g=cm3
1:79505r51:7960 g=cm3
1:79405r51:7950 g=cm3
1:79205r51:7940 g=cm3
1:79005r51:7920 g=cm3
1:78805r51:7900 g=cm3
Figure 2. Comparative crystal apparent density distribution for the
r51:7880 g=cm3
two RDX batches.
258 L. Borne, J. C. Patedoye, and C. Spyckerelle Propellants, Explosives, Pyrotechnics 24, 255Ä…259 (1999)
Table 3. Porous Volume (vp) for a Spherical Particle of Diameter D with only One Spherical
Cavity of Diameter d and Amount of Such Particles Needed to get 1 ml of Total Porous Volume
D d vp Amount (g) of particles with only one cavity
(mm) (mm) (ml) to get 1 ml of acetone if pores are saturated
with acetone
250 5 6.5E-08 225 g (15 278 875 particles)
250 10 5.2E-07 28.12 g (1 909 859 particles)
200 40 3.3E-05 0.22 g (29 842 particles)
250 40 3.3E-05 0.44 g (29 842 particles)
300 40 3.3E-05 0.76 g (29 842 particles)
250 80 2.7E-04 0.05 g (3 730 particles)
The amount of entrapped species (crystallization solvent performances of the sorting experiment in a disadvantageous
and water) were measured for crystals of most of the 11 range of apparent density. Between 1.7950 g=cm3 and
crystal classes sorted from each of the two RDX raw batches. 1.7980 g=cm3 the apparent density distribution curves of
The results are plotted on Figure 3a for RDX crystals the two raw batches have the highest slopes.
processed with acetone and on Figure 3b for RDX crystals For apparent density values below 1.7880 g=cm3, the slope
processed with cyclohexanone. The average crystal apparent of the apparent density distribution curve is reduced and the
density of the class is reported on the X axis. The amount of ef®cacy of the sort experiment is increased. A second sort on
solvent and the amount of water entrapped in the crystal are the class of crystals grown in cyclohexanone whose apparent
reported on the Y axis. A strong experimental correlation is density should be lower than 1.7880 g=cm3 shows that 93%
recorded between the crystal apparent density measurement in weight of crystals have an apparent density lower than
and the concentration of solvent and water entrapped in the 1.7880 g=cm3 and 99% have an apparent density lower than
crystals. These results demonstrate the ef®ciency and the 1.7900 g=cm3.
accuracy of the various measurement tools employed. The ef®ciency of the sort experiment is a function of the
ratio between the volume forces and the surface forces and
consequently is a function of the crystal size and the viscosity
of the sorting liquid mixture. The ef®ciency of the sort
4. Results and Discussion experiment decreases when the crystal size is reduced.
Practically below 100 mm the sort experiment needs to be
To check the accuracy of the sort experiment a second sort improved to reduce the role of the surface forces.
was performed with some of the previous crystal classes. The Nevertheless, the very good correlation between crystal
experimental results are compared with theoretical values apparent density and the solvent and water content is a
and plotted on Figures 4a and 4b. After the second sort some validation of the measurements.
crystals remain out of the apparent density limits resulting Evidence of air entrapped in the crystals has been proved
from the ®rst sort. as described before. Solubility of air in solvents may not be
For the selected class of crystals grown in acetone, the neglected. Part of the air released after dissolution of the
second sort shows that 44% in weight of crystals remain crystals may be solubilized by the solvent, that is why
between the apparent density bounds (1.7970 g=cm35r5 volumetric measurements were found to be not accurate
1:7980 g=cm3) and that 78% have an apparent density higher enough and may only be used for determination of orders
than 1.7960 g=cm3. The ®rst sort gave 47% in weight of raw of magnitude. Work is under progress to de®ne another
crystals whose apparent density is higher than 1.7960 g=cm3. technique for accurate measurement of air entrapped in
The ef®cacy of the sort is not 100%. These data illustrate the crystals.
Figure 3a. Trapped species in the various classes sorting from the Figure 3b. Trapped species in the various classes sorting from the
RDX lot using acetone as the recrystallization solvent. RDX lot using cyclohexanone as the recrystallization solvent.
Propellants, Explosives, Pyrotechnics 24, 255Ä…259 (1999) Quantitative Characterization of Internal Defects in RDX Crystals 259
Figure 4a. Accuracy of the sorting experiment (acetone). Figure 4b. Accuracy of the sorting experiment (cyclohexanone).
Symposium (International) on Detonation, Portland, August 28Ä…
Accurate measurements of the air content would allow
September 1, 1989, Oregon, pp. 18Ä…24.
calculations of the crystal apparent density and provide a
(2) A. C. Van Der Steen, H. J. Verbeek, and J. J. Meulenbrugge,
more complete checking of the apparent density measure-
``InŻuence of RDX Crystal Shape on the Shock Sensitivity of
ments. PBX's'', Proc. 9th Symposium (International) on Detonation,
Portland, August 28Ä…September 1, 1989, Oregon, pp. 83Ä…88.
(3) L. Borne, D. Fendeleur, and A. Beaucamp, ``Explosive Crystal
Properties and PBX's Sensitivity'', DEA 7304 Physics of
5. Conclusion
Explosives, Berchtesgaden, Germany, September 1997 and ISL
Report ISL=PU 358=97.
(4) L. Borne, ``InŻuence of Intragranular Cavities of RDX Particle
Experimental results and theoretical models underline the
Batches on the Sensitivity of Cast Wax Bonded Explosives'',
important role of internal crystal defects on the shock
Proc. 10th Symposium (International) on Detonation, Boston,
sensitivity of cast explosive formulations. This paper
July 12Ä…16, 1993, Massachusetts, pp. 286Ä…293.
(5) F. Baillou, J. M. Dartyge, C. Spyckerelle, and J. Mala, ``InŻuence
described accurate quantitative measurement tools to char-
of Crystal Defects on Sensitivity of Explosives'', Proc. 10th
acterize the internal defects in a lot of crystals. The measure-
Symposium (International) on Detonation, Boston, July 12Ä…16,
ment of the apparent density of the crystals provides
1993, Massachusetts, pp. 816Ä…823.
quantitative information on the global amount of internal (6) L. Borne, ``Microstructure Effect on the Shock Sensitivity of Cast
Á Á
Plastic Bonded Explosives'', Europyro 95 6ieme Congres Inter-
defects.
national de Pyrotechnie, Tours, 5Ä…9 juin, 1995, France, [Proc.]
This work shows that quantitative tools exist for the
pp. 125.
measurements of the chemical species entrapped in the
(7) P. M. Gallagher, M. P. Coffey, V. J. Krukonis, and W. W.
defects. A good correlation is found between the crystal Hillstrom, ``Gas Anti-Solvent Recrystallization of RDX: For-
mation of Ultra-®ne Particles of a Dif®cult-to-Comminute
apparent density measurements and the amount of entrapped
Explosive'', The Journal of Supercritical Fluids 5, 130Ä…142
species in the internal cavities.
(1992).
The detection of an internal cavity of 10 mm in a crystal
(8) M. Y. Lanzerotti, J. Autera, L. Borne, and J. Sharma, ``Crystal
whose average size is 200 mm is the low limit of the Growth of Energetic Materials during High Acceleration'',
Symposium Proceedings of the Materials Research Society,
measurements.
Boston, November 27Ä…30, 1995, Massachusetts, Vol. 418.
Theoretical models and computations suggest that the
(9) J. H. ter Horst, R. M. Geertman, A. E. van der Heijden, and G. M.
sizes of the internal crystal cavities could play a role on the
van Rosmalen, ``Bench Scale, Cooling Crystallization of RDX'',
Proceedings 27th International Annual Conference of ICT,
formulation sensitivity. Tests for convenient metrological
Karlsruhe, 1996, pp. 126=1Ä…126=13.
tools are in progress.
È
(10) U. Teipel, U. Forter-Barth, P. Gerber, and H. H. Krause, ``For-
Strong experimental correlation has been recorded between
mation of Particles of Explosives with Supercritical Fluids'',
the amount of internal defects and the shock formulation
Propellants, Explosives, Pyrotechnics 22, 165Ä…169 (1997).
(11) D. Spitzer and M. Samirant, ``Shock Solicitation of PETN Single
sensitivity(4,6). Accurate correlation between the formulation
Crystals Presenting Defects and Visualization of Hot Spots
sensitivity and the nature of the entrapped chemical species
Initiation'', Proc. 10th Symposium (International) on Detonation,
could now be performed taking care to control the other
Boston, July 12Ä…16, Massachusetts, 1993, pp. 831Ä…840.
parameters (amount of defects, crystal size, . . .).
Acknowledgements
This work was performed under SNPE contract with a govern-
mental (DRET) ®nancial support.
6. References
(1) H. Moulard, ``Particular Aspect of the Explosive Particle Size
Effect on Shock Sensitivity of Cast PBX Formulations'', Proc. 9th (Received December 4, 1998; Ms 32=98)