Combustion and Flame
Volume 14 Number 3 June 1970
The Combustion Reactions of a
Pyrotechnic White Smoke Composition
A. Jarvis
Technical Chemistry Division, Chemical Defence Establishment, Porton
Down, Salisbury, Wiltshire, England

[Scanned and you know what that means!!]

A common method for the production of a white screening smoke consists
of the pyrotechnic dispersion into the atmosphere of an aerosol of zinc
chloride. The pyrotechnic mixtures employed for this purpose commonly
contain zinc oxide, calcium silicide, and a chlorine-donating material such
as hexachloroethane. The range of mixtures of this type currently
employed in British smoke generators is zinc oxide 47%-29%, calcium
silicide 7%-25%, and hexachloroethane 46%, higher proportions of the
fuel, calcium silicide, producing faster-burning compositions. The pyrolysis
of hexachloroethane and the reactions occurring between the principal
constituents of this mixture, taken in pairs and as a whole, have been
elucidated by the technique of differential thermal analysis.
Hexachloroethane is pyrolyzed to carbon and chlorine via the intermediate
formation of carbon tetrachloride, tetrachloroethylene, and
hexachlorobenzene. Calcium silicide reacts with hcxachloroethane in four
stages to form calcium chloride, carbon, and silicon, reaction intermediates
in this process being identical with those in the hexachloroethane
pyrolysis. Zinc oxide also reacts in a stepwise manner with
hexachloroethane, finally producing zinc chloride, carbon monoxide, and
carbon dioxide; reaction intermediates in this process are carbon
tetrachloride, tetrachloroethylene, phosgene, and a zinc oxychloride. The
temperature of initial reaction in the ternary system is 50o-80o below that
observed for the zinc oxide-hexachloroethane or the calcium
silicide-hexachloroethane mixture, because of interaction of the two binary
reaction schemes.

Introduction
Previous knowledge of the behavior of pyrotechnic white smoke
compositions of the type zinc oxide-calcium silicide-hexachloroethane
(HCE) has been empirical because of.the technical difficulties of studying
vigorous pyrotechnic reactions by classical methods. The development of
modern techniques such as differential thermal analysis (DTA) [1] has ren-
dered it possible to undertake a study of the complex reactions which
occur in pyrotechnic mixtures.
The range of mixtures of the above type currently employed in British
smoke generators for the production of zinc chloride smokes is as follows:
zinc oxide 47%-29%, calcium silicide 7%-25%, and hexachloroethane
46%. The temperature of burning of this mixture lies in the range from
700oC (low proportion of the fuel, calcium silicide) to 1100oC (high fuel
content), elevation of the temperature resulting in a proportional increase in
the rate of burning of the pyrotechnic mixture.

This paper deals with the pyrolysis of HCE and the identification of the
reactions which occur between the principal constituents of this white
smoke composition, taken in pairs and as a whole. The study is based on
the use of DTA.

Throughout this paper calcium silicide will be formulated as CaSi3, this
corresponding to the approximate atomic ratio of its constituents.

Experimental Apparatus, Materials, and Procedures
The apparatus consisted of a Stanton TR-01 thermobalance modified to
perform simultaneous thermogravimetry and DTA [2]. However, in this
work only DTA was used since it was necessary to employ a sealed
reaction system. The DTA measuring head was an alumina block, of 20
mm diameter and 13 mm depth, fitted to the top of the thermobalance rise
rod. The block had two wells (each of 6.5 mm diameter and 10 mm depth)
to accommodate two dimped crucibles and was supplied with a fourbore
rod and two opposed Pt/Pt-Rh (13%) thermocouples. The crucibles used
were 4-mlcapacity quartz cells (4 mm internal diameter) having a dimped
base which seated over the measuring thermocouples. All the work re-
ported in this paper was performed in these sealed cells since the HCE
would otherwise have been lost by sublimation before reaction occurred
with the other components of the composition.

Both the ambient and the differential temperatures (T and deltaT,
respectively) were recorded simultaneously. The ambient temperature was
measured by the thermocouple below the reference cell.
The materials used were closed-sieved in the following ranges to reduce
second-order effects due to particle size differences: HCE 22-25 BS sieve;
zinc oxide, 100-120 BS sieve; calcium silicide, 240-300 BS sieve. These
size ranges approximate those normally found f or these materials in British
smoke generators. Each component was dried and stored in a vacuum
desiccator before use.

Because of the practical impossibility of ensuring that very small samples
are representative and consistent in composition when the original mixture
contains constituents differing widely in particle size and density, the DTA
samples incorporating HCE were prepared by the method of separate
addition, that is, this component was added separately to the sample cell
after the other component(s). Since the reactions of interest occurred
between the gaseous organic and solid inorganic materials, the method of
separate addition was justified; this technique also resulted in a more
stable base line, since the packing of the inorganic component around the
thermocouple dimp was less disturbed by the evaporation of the HCE than
was the case for an intimate mix. The samples were packed down by
tapping the cell on the bench .

The reference material used was calcined alumina of the same weight as
the sample under study. The weights of the sample constituents of each
DTA experiment were taken in the same relative proportions as those
outlined above.

The products of the reactions occurring between the various components
were identified by gas-liquid chromatography and infrared spectroscopy.

Results

Analytical
Microanalysis of the HCE gave 89.45% (theoretical 89.58%) chlorine, and
gas-liquid chromatography showed only one peak.
Classical chemical and spectroscopic analysis showed the calcium
silicide to contain 32% calcium, 62% silicon, 3% iron, and 2% aluminium,
together with traces of titanium, manganese, magnesium, barium, and
strontium. Thus this material appears from the calcium-silicon phase
diagram [3] to be mainly a mixture of silicon and calcium disilicide (the
approximate atomic ratio being Ca:3Si).

Differential Therma l Analysis
1. Hexachloroethane

The sample weight taken was 45.8 mg, and the mean heating rate was
4.1oC/min. The changes giving rise to the peaks in Table I can be
summarized as follows:

C2Cl6 (rhombic)  ?  C2Cl6 (triclinic)     A (51oC)
C2Cl6 (triclinic)    ?  C2Cl6 (cubic)       B (75oC)
C2Cl6 (cubic)        ? C2Cl6 (liquid)       C (188oC)
C2Cl6 (liquid)        ? C2Cl1 (gas)          D (278oC)
2C2Cl6        ? 2CCI, + C2Cl4
2CCl4        ? C2Cl4 + 2Cl2        E (728oC)
  3C2Cl4         ?  C6Cl6 + 3Cl2
C6Cl6        & nbsp;  ?  6C + 3CI2              F (830oC)
2C+02         ? 2CO                      G (874OC)

The peak temperature for each change is given. Accepted values for
transitions A, B, and C are 45', 72', and 187C, respectively. No reference
to phase change D has been found, but analysis of the cell contents after
peak D showed them to be pure HCE.

Peaks E-G were characterized by stopping the DTA run before and
immediately after each peak, quenching the sample cell in cold water, and
analyzing the contents. The initial pyrolysis of HCE to carbon tetrachloride
and tetrachloroethylene has also been postulated by Dainton and Ivin [4].
2. Zinc Oxide (Analar)

When 49.2 mg of zinc oxide was heated from room temperature to 900 oC
at a mean heating rate of 4.0oC/min, no peaks were observed in the
differential thermogram.

3. Calcium Silicide
At a mean heating rate of 4.3oC/min from 20oC to 900oC, 49.6 mg of
calcium silicide appeared to be thermally stable since no DTA peaks were
observed.

4. HCE + Calcium Silicide
Initially 46 mg of HCE and 7-25 mg of calcium silicide were heated at a
nominal rate of 4oC/min (Table 2). The four HCE transitions were observed
at the expected temperatures, and the exothermic peaks thus
corresponded to reaction of the solid inorganic with the gaseous organic
component.

It was found that reaction proceeded in several stages since two sets of
peaks were observed. As the calcium silicide content of the mixture
increased, the first set of exothermic peaks was poorly resolved into a
triplet (Table 2) at 372'-375'C. These peak temperatures were little affected
by changing the HCE/calcium silicide ratio. The peak a rea, however,
showed a definite reduction as the calcium silicide content increased. This
is to be expected from the work of Sewell and Honeyborne [5], who
showed that peak area is directly proportional to the heat of reaction per
unit mass of reacting material multiplied by the mass fi-action of this
material in the sample.

Following the multiplet exotherm was a single exothermic peak. This
increased in size and appeared at successively lower temperatures as the
proportion of calcium silicide increased.
Optimum resolution of the multiplet exotherm. was achieved when a 20-mg
HCE- 12 mg calcium silicide sample was used (Fig. 1). The exothermic
peaks were characterized by analysis of the products of reaction after each
peak, thus:

CaSi3 + 3C2Cl6  ?  CaCl2 + 2CCl4 + 2C2Cl4 + 3Si    A (381oC)
2CaSi3 + 2CCl4 ?    2CaC12 + C2Cl4 + 6Si           ;        B (387-C)
3CaSi3 + 3C2Cl4 ?  3CaCl2 + C6Cl6 + 9Si                   C (398-C)
3CaSi3 + C6Cl6   ? 3CaCl2 + 6C + 9Si                        D (472-C)

Support for the above scheme was furnished by heating the mixture to
900'C when the DTA record exhibited a sharp endotherm at 772oC, in
agreement with the accepted value for the fusion of calcium chloride.
Furthermore, DTA of calcium silicide mixed with each of the above reaction
intermediates resulted in the carbon tetrachloride mixture giving three
exotherms (at 345o, 349o, and 402oC),  tetrachloroethylene two (at 396o
and 516oC, and hexachlorobenzene one (at 463oC.

The silicon moiety of the calcium silicide was thought to rea ct little if at all
with HCE (or its decomposition products) since DTA of this element and
chlorinated hydrocarbons resulted in a small broad exotherm (slow
reaction at about 380oC.)
5. HCE + Zinc Oxide (Analar)

When 46 mg of HCE was heated with 47-mg to 29-mg quantities of zinc
oxide at a nominal heating rate of 4'C/min, the usual HCE transitions were
observed. Hence the reactions of these two compounds were of the
gas-solid type.
It was found that the reaction peaks were not significantly affected (with
regard to peak area or temperature) by variation of the zinc oxide content
of the mixture. Figure 2 is therefore representative of all the mixtures used
(see Table 3). The DTA peaks in Fig. 2 are thought to be due to the
following reactions:

2C2Cl6  ?  2CCl4 + C2Cl4
6ZnO + 2CC14  ? 2(2ZnO.ZnCl2) + 2COCl2          A (348oC)
6ZnO + 2COC12 ? 2(2ZnO.ZnCl2) + 2CO2< BR>Not fully elucidated                               B  (443oC)
Pyrolysis of excess chlorinated hydrocarbons
(see Section 1)                              C (686oC

The reaction intermediates were identified after each peak, those after
peak A being mainly HCE, tetrachloroethylene, and carbon dioxide,
together with small amounts of carbon tetrachloride and phosgene;
substantial amounts of zinc oxide also remained after this first reaction
sequence. It seems probable therefore that some stable zinc oxychloride
of empirical formula 2ZnO - ZnG2 is formed. The singlet nature of peak A
is probably due to the comparatively low temperature of the reactions of
carbon tetrachloride (224'Q and phosgene (277'Q with zinc oxide as
compared with the temperature of the reaction sequence shown above.
The reaction(s) responsible for exotherm B cannot be ascribed with any
certainty from the available results. There exist three possibilities, these
being (1) the zinc oxychloride reacts with tetrachloroethylene thus:

2ZnO - ZnCl2 + C2Cl2 --+ 3ZnCl2 + 2CO

(DTA of zinc oxide + tetrachloroethylene gives the above products at a
peak temperature of 533oC, or (2) the oxychloride reacts with excess HCE
by a scheme similar to that for peak A (Fig. 2), or (3) the first and second
mechanisms could proceed concurrently. The second of these possibilities
seems the least likely at the present time since after peak B the amount of tetrachloroethylene in the
sample cell was low and a substantial amount of carbon monoxide was
found.

Finally, it should be noted that the reaction of HCE with zinc oxide is
much less exothermic than with calcium silicide.

6. Zinc Oxide + Calcium Silicide
Mixtures of these two substances heated to 90oC at a nominal rate of
4oC/min were found to be chemically inert.

7. Zinc Oxide + Calcium Chloride
This binary system was studied since DTA of a complete pyrotechnic white
smoke composition (see Section 8) did not show a calcium chloride fusion
endotherm. as expected. The two components were taken in
equirnolecular quantities (total weight 50 mg) and heated at a nominal
4oC/min in a sealed quartz cell. The only DTA peak observed was the
calcium chloride melting endotherm at 771oC.

8. HCE + Zinc Oxide + Calcium Silicide

These components were examined in various ratios corresponding to the
smoke composition in current use, sample weights being 100 mg total. At a
nominal heating rate of 4oC/min the usual HCE transitions were observed
(peak A, Fig. 3, being the vaporization endotherm) be fore reaction of this
compound with the inorganic solids.

The following reaction scheme is believed to account for exothermic
peaks B-E in Fig. 3 and Table 4:

Combustion of a Pyrotechnic White Smoke Composition
CaSi3 + 3C2Cl6 CaCl2 + 2CC14 + 2C2Cl4 + 3Si
6ZnO + 2CCl4  ? 2(2ZnO-ZnCl2) + 2COCl2         B
6ZnO + 2COCl2  ?  2(2ZnO-ZnCl2) + 2CO2
3CaSi3 + 3C2Cl4 ?  3CaCl2 + C6Cl6 + 9Si          C
       2ZnO-ZnCl2 + C2Cl4  ? 3ZnCl2 + 2CO                        D
3CaSi3 + C6Cl6 ? 3CaCl2 + 6C + 9Si                     E

Thus, on stopping the DTA run after exotherm B, the volatile produc ts
were found to be mainly tetrachloroethylene, carbon dioxide, and some
HCE. It may be noted that in these ternary mixtures exotherm B occurs at
50'-80'C below the lowest exothermic peak temperatures found in any of
the binary mixtures. This is probably due to the extremely rapid rate of
reaction of zinc oxide with carbon tetrachloride (see Section 5) as the latter
is initially produced at about 300'C as shown above; the singlet nature of
peak B is similarly explained. As the calcium silicide content of the mixture
decreases, peak B is reduced in size, since less carbon tetrachloride is
available for reaction with the zinc oxide.

After peak C the organic products were found to be mainly
hexachlorobenzene together with some tetrachloroethylene, the latter
probably being produced from any excess HCE remaining after reaction
sequence B. Exothermic peak C is seen to decrease rapidly and finally
disappear on reducing the calcium sili cide content of the ternary mixture.
Also, peak E decreases similarly since reaction C is the source of the
hexachlorobenzene reactant for reaction E.

The exothermic reaction giving rise to peak D produced substantial
quantities of carbon monoxide, indicating the reaction of zinc oxychloride .
(or oxide) with tetrachloroethylene. Peak D becomes more pronounced as
the calcium silicide content of the mixture decreases, since more
tetrachloroethylene becomes available for reaction by virtue of reaction C
being diminished.

Finally, on heating all the ternary mixtures to 900oC no calcium chloride
melting endotherm. was produced. Since this compound is produced by
reactions B, C, and E and does not react with zinc oxide (Section 7), the
absence of this endotherm is thought to be due to the dissolution of the
calcium chloride in the molten zinc chloride produced by reaction D.

Practical Considerations

The reaction sequenc e outlined in Section 8 indicates that the combustion
products of this type of white smoke composition may contain phosgene
and carbon monoxide. However, on burning a sample and examining the
volatile products by gas-liquid chromatography, the concentrations of these
two species were found to be extremely low. This is due to the phosgene
being produced at a higher temperature than that needed for reaction with
zinc oxide; consequently the phosgene appears as only a transient
intermediate. Also, carbon monoxide is a product of the
tetrachloroethylene-zinc oxide reaction, which occurs at a higher
temperature than the tetrachloroethylene-calcium silicide reaction, making
the latter the preferred mode of decomposition for this chlorinated species.
Gas-liquid chromatography of the volatile products of the burning
composition shows the major constituent to be tetrachloroethylene,
indicating that the reaction of this material with even calcium silicide is only
minimal in practice and that the major combustion and smoke-producing
reactions are those in reaction sequence B of Section 8. Thus smoke
production is directly related to the amounts of fuel (calcium silicide) and
smoke producer (zinc oxide) present in the pyrotechnic composition.

Conclusions

Hexachloroethane reacts with calcium silicide in four stages, the first of
which- produces tetrachloroethylene and carbon tetrachloride. The latter
reacts very rapidly with zinc oxide, forming an oxychloride and phosgene,
which then similarly produces an oxychloride and carbon dioxide. This
series of reactions is selfreinforcing with a consequent lowering of the
ignition temperature. The temperature of reaction between phosgene and
zinc oxide is so low (ca. 280'Q, compared with the temperature of the
burning composition, that the phosgene concentration in a smoke cloud is
very low.

Further reactions can occur be tween calcium silicide or zinc oxychloride
and tetrachloroethylene, producing hexachlorobenzene and calcium
chloride or carbon monoxide and zinc chloride, respectively. Finally, the
hexachlorobenzene can react further with calcium silicide to give the metal
chloride, carbon, and silicon. These reactions occur to only a limited extent
in a burning composition, however, and the amount of carbon monoxide
produced is very small.

The ease and the extent of smoke formation are directly dependent on
both the fuel (calcium silicide) and the smoke producer (zinc oxide) content
of the pyrotechnic mixture.

Technical assistance in this work was given by Mr. C. Wort. The
spectroscopic analysis of calcium silicide was performed by Mr. R. A.
Mostyn of D.C.I. Woolwich Arsenal.

References
I . GARN, PAUL D., Thermoanalytical Methods of Investigation. Academic:
New York (1965).
2. Stanton Instruments Technical Information Sheet 4.
3. SMITHELS, C. J., Metals Reference Book, Vol. 2. Butterworths: London.
4. DAINTON, F. S., and IVIN, K. J., Trans Faraday Soc., 46, 295(1950).
5. SEWELL, E. C., and HONEYBORNE, D. B., The Differential Thermal
Analysis of Clays (ed. R. C. MacKenzie), Chap. 111. Mineralogical
Society, London (1957).
(Received September, 1969; revised January, 1970)
Table 1. DTA of Hexachloroethane (45.8 mg)
Figure 1. IDTA of HCE (19.8 mg) + CaS'3 (12.0 mg). Peak
temperatures ('C): A 381', B 387', C 398', D 472'.
* Pt/Pt-Rh (130/0) thermocouple.
Table 2. DTA of Hexachloroethane + Calcium Silicide-, "
Table 3. DTA of Hexachloroethane + Zinc Oxide—
Figure 2. DTA of HCE (45.9 mg) + ZnO (39.0 mg).
Peak temperatures ('C):JO A 348', B 443', C 6861. * Pt/Pt-Rh
(130/0) thermocouple.
Figure 3. DTA of HCE (46.0 mg) + ZnO (34 mg) + CaS'3 (20 mg). Peak
temperatures ff): A 266', B 287', C 419', D 477% E 510o. *Pt/Pt-Rh (13%)
thermocouple.
Table 4. DTA of Hexachloroethane + Calcium Silicide + Zinc Oxide