THE SENSITIVITY OF INITIATOR EXPLOSIVES TO

MECHANICAL IMPACTL
BY N. R. S. HOLLIES,~
N. R. LEGGE,~
AND JOHN L. MORRISON

ABSTRACT
An original ball-drop impact apparatus has bee11 consiructed and used to test
the sensitivities of dextrin lead azide and mercury fulminate to mechanical
Impact. Ball masses artd heights were both varied. Both the net kinetic energy
and the change of nlomentuIn gave c o n t i n ~ ~ o ufunctions
s with the percentage of
detonations. Con~parisonsof the absolute values of the net kinetic energy and
the change of momentum with another reported in the literature suggest t h a t
momentum is the more important factor in determining the probability of
detonation. Values for the momenturn a t 50% detonatior~sfor lead azide and
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

mercury fulminate are 4.6 X 10' and 2.6 X lo4 c.g.s. units, respectively, with
an area of contact of about 0.08 cm.? The corresponding times of impact were
found to be ahout 2.2 X lo-' and 1.9 X 10-"ec.

The question of whether the probability of the detonation of an initiator

explosive is chiefly determined by the kinetic energy or the momentum of the
impacting mass is of theoretical interest in interpreting the process of detonation
by impact (8). In their earlier work, Taylor and Weale (9) suggest that kinetic
energy is the important factor, although in 1938 (10) they introduce experiments
and calculations based on momentum. Taylor and Weale do not mention the
rebound of the steel balls used in their impact tests, although the measurement
of rebound was introduced decades ago (referred to in Ref. 11). Powell, Skelly,
For personal use only.

and Ubbelohde (8) measured the momenta and kinetic energies of impact for
mercury fulminate and lead styphnate. In their calculations, they assumed
without measurement a definite value for the coefficient of restitution of the
impacting mass. In the present work, some measurements of the net kinetic
energy, change of momentum (hereinafter called impulse), and time of impact
for the initiator explosives mercury fulminate and lead azide are reported. The
time measurements allowed estimations to be made of the forces involved and
these were found to be comparable with those found by Taylor and Weale (10)
by an entirely independent method.
EXPERIMENTAL
Materials
Dextrin lead azide was prepared from sodium azide which had been recrystal-
lized from a water-acetone mixture. Separate 50 ml. aqueous solutions containing
4 gm. sodium azide and 10.2 gm. lead nitrate C.P. were simultaneously added
from burettes (delivery time-134 sec.) into 250 ml. of 0.5% aqueous solution
of dextrin. Microscopic examination showed that the crystals of lead azide were
of the same form as those prepared by Lowndes (6).
1 Manziscript received Fcbruary $4, 1953.
Contribzilion from the Departnsent of Chemistry, University of Alberta, Edtnonton, Alta.
2 Graduale student, 1944. Present address: Harris Research Laboratories, Washington, D.C.
3 Graduate slzident, 19/t$-/tS. Present address: Kentucky Rubber Company, Loziiseille, K y .
746
 HOLLIES E;T AL.: SEh7SITII'ITY OF I N I T I A T O R EXPLOSII'ES 747
Mercury fulminate was re pared by a method given by Davis (5, p. 406).
Impact Apparatus
The essential parts of the impact apparatus are shown in Fig. 1. The hammer
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

F I G . I. Impact apparatus (schematic).

A-hammer D-s tee1 ball G-steel ring for static load
B-base plate E--hardwood guide . H-bolt hole
C-cavity in base plate F-vent to suction 1-1 beam
For personal use only.

was made from the rear axle of a 1929 model Pontiac automobile. I t rested in a
concentric cavity in a 0.5 in. thick steel base plate which had been cut from a
high carbon steel road grader blade. The steel base plate was carefully faced
and fastened securely by six large bolts to the smoothed upper face of an 18 in.
I beam (approx. wt. = 110 Ib.). Both the hammer and base plate were hardened
by heat treatment and quenching. This arrangement of base plate and I beam
ensured a large mass as well as the possibility of hardening the anvil. In all the
experiments the hammer was under a static load obtained by means of a second
class lever. The total load, including the hammer of 211 gm., was 40,098 gm.
Steel balls (S.1C.F. ball bearings) of various sizes were dropped on the hammer
by release from an electromagnet and their rebound was measured by the shadow
cast on a lighted calibrated scale.
The original cavity in the base was obtained by bouncing a 1 3 in. diameter
ball (larger than any used in the impact tests) on the hammer. Then, in carrying
out a series of tests, the smallest balls and heights of fall were used a t first,
and these were increased in succession. This procedure prolonged the good
condition of the base plate cavity and of the hammer. Whenever the end of the
hammer became pitted by explosions, it was reshaped on a lathe to a radius of
curvature of 5/32 in. with successive emery paper sizes 2, 00, and 0000. This
procedure was necessary after 15 to 20 detonations of lead azide. Then the
cavity in the base plate was repounded until the normal bounce height was
secured.
 748 C A N A D I A N JOURNAL OF CHEBfISTRY. VOL. 31

In carrying out an impact test, the cavity in the base plate was filled with
initiator explosive by dusting with a camel's hair brush. (The initiator waskept
over calcium chloride between tests.) The hammer and weight arm were gently
lowered on top of the sample and the extra weight was then placed on the weiiht
arm.
There was no difficulty in detecting the occurrence of an explosion, although
mercury fulminate exhibited the phenomenon of partials a few times.
Time of Impact Afeasurements
The time of impact was measured in a way similar to t,hat used by Bowden
and Tabor (4). A wave controller using a cathode ray oscilloscope was designed
and built by Dr. H. E. Johns* and one of us (N.R.S.H.). The controller used a
thyratron which was applied to the horizontal plates to produce a linear beam of
definite periodicity across the oscilloscope screen. At the same time a transient,
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

which was produced by the closing of a voltage circuit by the impact of the
hammer on the base plate, was applied to the vertical plates. The resulting
impact wave trace on the oscilloscope screen was photographed with Kodak
Verichrome film. The period of the trace was calibrated by photographing the
trace of a standard 10,000 or 20,000 cycle set.-' wave on the same film.
Because of a poor estimate of the expected length of the wave trace, the
firing mechanism was inaccurately timed, and the length of the transient was
estimated to be as much as 25% too short; that is, the times listed in Table I11
are a t the most about 25% low; however, they are still related relatively.
Bases of the Calculations
For personal use only.

The percentage of detonations are compared with (a) the kinetic energy
assuming no rebound of the ball, (b) the net kinetic energy, and (c) the impulse
or change in momentum. These aregiven by the following equations, respectively:
( a ) mgh,
( b ) mg(hl - h2) = 980 mhl(1 - e",
(c) Ft = m(vl - v2) = 44.3 m(1 e)h$, +
where m = ball mass, g = gravitational constant, hl = initial fall height,
hz = rebound height, e = uz/vl = (h2/h1)+= coefficient of restitution, F =
force, t = time over which ball goes from v l to v2, v l = falling velocity before
duration of contact, vz = rebounding velocity after duration of contact.
RESULTS A N D CALCULATIONS
Impact Tests
The results of about 500 impact tests on dextrin lead azide and 300 on mercury
fulminate are given in Tables I and I1 respectively. At least 20 and sometimes
100 trials were made for each ball a t each height (the numbers are given in
brackets after the percentage detonations). In Figs. 2 and 3 the percentage
detonations are plotted against gross kinetic energy, net kinetic energy, and
impulse respectively.
Time of Impact
The times of impact as measured under various conditions are given in
* A t the time, Assistant Professor, Deparfmet~tof Physics, University of Alberta.
 H O L L I E S E T AL.: S E X S I T I V I T Y O F I N I T I A T O R EXPLOSII'ES 749

TABLE I

::I : :
di?m.,
\

mass,
SENSITIVITY
I
OF LEAD

Height
(hl),
cm.
I 1
Detonations,
%
I

mgi,
ergs X
~
AZIDE TO KINETIC ENERGY .AND IMPULSE

I
Net kinetic
energy,
ergs X
I

Impulse,
c.g.s. units X 1 0 - 4
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

TABLE I1
SENSITIVITY
OF MERCURY FULMINATE TO KINETIC ENERGY AND IMPULSE

Ball Ball Height Net kinetic

diam., mass, (hl), r ~ e t o ~ t i o n sergs X
, l mgh, energy, Impulse,
in. gm. cm. ergs X c.g.s. units X 1 0 - 4
For personal use only.

F I G . 2. Compariso~lsof the percentage of detonations of destrin lead azide with gross

and net kinetic energies and impulse.
 CANADIAN J O U R N A L OF C I I E d f I S T R Y . VOL. 31
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

FIG.3. Con~parisonsof the percentage of detonations of mercury fulminate with gross

and net kinetic energies a n d impulse.

Table 111. Between the tests on lead azide and mercury fulminate the base
plate was reground and rehardened. This accounts for the slight change in the
For personal use only.

coefficient of restitution e of the 7 / 8 in. ball in Tables I and 11. Also, in some
of the time experiments, norite was placed in the cavity between the hammer
and base plate. The use of norite probably simulated the conditions of impact
when initiator crystals rested between the hammer and base.
Hertz's theory of colliding metal pieces (1) shows that t(rv/m2)lI5= k,
a constant, where t = time of impact, r = radius of steel ball, v = vl = velocity
before impact, m = ball mass. The constant k has been calculated for the norite
time experiments and is included in Table 111. The evaluation of k permits an
estimation to be made of the time of impact within the lead azide detonation
range where norite experiments were not made.
Calculation of Impulsive Pressures
Assuming that impulse determines the probability of detonation, an estimate
of the impulsive pressure or force per unit area a t 50y0 detonations can be made.
The apparent area of contact between the hammer and base plate was Oil0
sq. cm. The hammer was under a static load of 400 kgm. cm.-Wnder this
compression, the 'true' area of contact with the initiators can be calculated
from their crystal densities a t known compressions (7). These are 0.88 of the
apparent area for mercury fulminate, and 0.75 for lead azide.
For Mercury Fulrninate
Total impulse a t 50y0 detonations = 2.6 X l o 4 c.g.s. units. Time of impact a t
!joy0detonations = 1.9 X sec. Force = I / t = 1.37 X lo8dynes. Impulsive
 HOLLIES E T .4iA.: S E N S I T l 1 7 T Y OF I N I T I A T O R EXPLOSIVES

TABLE I I I

Ball $ism.,
~n.
Height ( k , ) ,
cnl. / Time of impact,
s e c x lo4
Original base plate
I I

R q r o z ~ n dbase plate
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

I I

Regrou~zdbase plate wirh norile / k (Hertz) X lo4

I I I '
For personal use only.

pressure = 1.37 X..lD8/0.88 X 0.1 = 1.56 X l o Y dyne ~ m . -or~ 11.3 ton in.-?
For Lead Azide
Total impulse a t 50% detonations = 4.6 X l o 4c.g.s. units. Time of impact a t
50% detonations (based on norite times and Hertz's formula) = 2.2 X 10-4
sec. Impulsive pressure = 2.1 X 108/0.75 X 0.1 = 2.8 X l o y dyne ~ m . - or~
20.3 ton in.-"
(Note: As the time of impact may be about 25% low, the above impulsive
pressures may be about 25% high.)
'Coeficient of Restitution
Values of the coefficient of restitution e are included in Tables I and II.
Their measurement is a useful method of testing the stability of the impact
apparatus. The constancy of e for any one ball over the range of fall heights
used is considered an indication of a rigid system (1). The values of e were
found to be completely independent of the presence or absence of initiator
explosive crystals, or of the occurrence or nonoccurrence of,detonations.
The relation between ball weights and e is given in Fig. 4. Shown are the
results for (a) an earlier, more complex, apparatus which included a ball bearing
 CANADIAN JOURNAL OF CHEAIISTRY. 1'OL. 31
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

50 100 150 200 2 50

BALL WEIGHT, gm.
FIG.4. The relationship between coefficient of restitution and ball weight for various
assentblies.

as part of the hammer assembly, (b) the present apparatus (Fig. l ) , and ( c ) the
same balls when dropped directly on the base plate. I t would appear that the
For personal use only.

curious periodic relation between ball weight and e arises largely from the
hammer part of the apparatus, and may be related to its resonating frequency.
DISCUSSION
Although the range of fall heights used in the present work was not sufficient
to give overlapping in the critical regions of percentage of detonations (Figs. 2
and 3), yet the general trend is quite clear. I t is apparent that both the net
kinetic energy and the impulse are continuous functions of the percentage of
detonations. Thus, no discrimination between the relative importance of kinetic
energy and impulse in determining the probability of detonation can be made
merely by varying the size of balls. If this is true for all impact machines, then
one must look in other directions for such a discrimination.
Kinetic energy is dissipated in many directions, whereas impulse is probably
conserved in one direction. A comparison of the absolute magnitudes of the net
kinetic energy and the impulse for different impact assemblies may provide
the basis for a critical test of which is the more important factor in detonation.
In the present case, the assembly could have been varied by altering the mass
of the hammer-this was not attempted.
Powell, Skelly, and Ubbelohde (8) have also determined the variation of
the percentage of detonations with kinetic energy and impulse for mercury
fulminate. Their value for the net kinetic energy a t soy0 detonations was
about 8 X lo3 ergs, compared with the present result of 1.7 X lo6 ergs. On the
other hand, their value for the impulse a t soy0 detonations was 1.2 X lo4
 H O L L I E S E T :1L.: S E N S I T I V I T Y O F I N I T I A T O R E X P L O S I I . E S 753

c.g.s. units, which is quite comparable with the present value of 2.6 X l o 4 c.g.s.
units on the following grounds. Their area of contact appears to be about one
half of the present one; their drift, containing a 3/16 in. diameter ball bearing,
rested on a flat brass disk which was resting in turn on a prepressed sample of
the initiator explosive; in the present case, the end of the hammer is the equiva-
lent of a 5/16 in. diameter ball, and it rests on a sample of the initiator which is
resting in turn in a cavity concentric with the end of the hammer. Thus, it
would appear on the basis of the above comparisons involving radically different
impact machines that impulse is the more important factor in initiating detona-
tion.
In their calculations, Powell et al. assumed a value of 0.9 for the coefficient
of restitution. Moreover, their impact assembly was not rigid: "a fraction of
the momentum of the drift is lost owing to imperfect rigidity of the supports for
the initiator (i.e. the brass disks, the Hoffman roller, etc.)"-(8, p. 294). I t is
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

possible that their results would have been more consistent if they had measured
e and used a more rigid assembly. In spite of this, the impulses for two different
drifts are parallel to each other over the range of percentage of detonations,
and only differ by about 20y0 in the case of mercury fulminate and 12y0 in the
case of lead styphnate.
A hypothesis which favors kinetic energy is that of Bowden and Gurton (2)
who, on the basis of work with added grits, come to the conclusion that the
initiation of explosives by mechanical impact is a frictional phenomenon.
In their words: "The mechanical energy of the blow must first be degraded t o
heat to form a hot spot of small but finite size. Thermal initiation then occurs
For personal use only.

a t this hot spot." This round-about transfer of energy (i.e. a transfer from
mechanical to thermal to chemical energies) is difficult to conceive on thermo-
dynamic grounds. The more direct transfer from mechanical to chemical
energies would be suggested if impulse is the factor determining the probability
of detonation on impact.
In another paper (3), Bowden and Gurton report what they designate as
"the delay between impact and explosion" times measured by means of the high
speed photography of actual explosions on impact. Practical difficulties prevented
the measurement of the times for ordinary lead azide and mercury fulminate
but in the presence of a glass grit particle, they were of the order of 1 X sec.
This is very similar to the times of impact reported in the present work, and
would suggest that the so-called 'delay7-presumably to allow mechanical
energy to be changed to thermal energy and then to detonation-may be none
other than the normal time of impact connected with the direct transfer of
momentum. For larger crystals of lead azide, Bowden and Gurton fouild the
'delay' time to be about 1 X sec., an observation difficult to accourlt for
on the basis of their hypothesis, but not so on the basis of the momentum
hypothesis.
The magnitude of the impulsive pressures necessary for the detonation of
mercury fulminate may be compared with another estimate available in the
literature. Taylor and Weale (10) compared the impulse necessary to detonate
mercury fulminate with that necessary to indent a lead block. They concluded
 754 CANADIAN J O U R N A L OF CHEIZZSTRY. VOL. 31

that "the impulsive stress required to initiate a thin layer of mercury fulminate
was not less than 7 tons per sq. in." (10, p. 999). Further, they estimated the
time of impact to be of the order of sec.
Some realization of the apparatus difficulties involved in this type of work
may be seen by an examination of the velocities of impact necessary for the
detonation of initiator explosives. Referring to the work of Andrews (1) and of
Bowden and Tabor (4), it is apparent that the velocities used here (343 to 486
cnl. sec.-l) are well beyond the velocity where the plastic deformation of steel
sets in. Bowden and Tabor found that for very hard tool steel a t higher velocities
(100 to 200 cm. sec.-l) the conductance curves for the ,time of impact became
asymmetric, showing that plastic deformation occurs. Andrews has shown that
Hertz's well-known formula only strictly applies for pure elastic rebound,
which only occurs a t low velocities of the order of a few cm. set.-I I t would
appear, however, from the present results, that the occurrence of some plastic
Can. J. Chem. Downloaded from cdnsciencepub.com by 103.255.6.98 on 02/17/21

deformation is not too important ~ r o v i d e dthat an otherwise rigid assembly is

used in impact tests.
ACKNOWLEDGMENTS
The authors acknowledge financial assistance from the National Research
Council, certain technical assistance by Drs. H. E. Johns and I<. A. Clark, and
helpful discussions with Prof. F. Keeping and Dr. D. B. Scott.
REFERENCES
1. ANDREWS, J . P. Phil. Mag. 9: 593. 1930.
2. BOWDEN, F. P. and GUKTON, 0. -4. Proc. Roy. Soc. (London), A, 198: 337. 1949.
0 . A. Proc. ROY.SOC.(London). ,. A.. 198: 350. 1949.
For personal use only.

3. BOWDEN. F. P. and GURTON.

4. TABOR.D. Engineer, 172: 380. '1941.
BOWDEN: F. P . ~ n d
5. DAVIS,T. L. I he chemistry of powcler and explosives, Vol. 11. John Wiley and Sons,
Inc., New York. 19-13,
6. LOWNDES, A. G. Trans. Faraday Soc. 16: 128. 1921.
7. MURAOUR. H. Bull. soc. chim. France. 51: 1152. 1932.
8. POWELL,W. J., SKELLY,H., and UBBELOHDE, A. Ti. Trans. Roy. Soc. (London), A,
241: 287. 1949.
9. TAYLOR, W. and WEALE,A. Proc. Roy. Soc. (London), A, 138: 92. 1932.
10, TAYLOR, W. and WEALE,A. Trans. Faraday Soc. 34: 995. 1938.
11. URBANSKI, T. 2. ges. Schiess- LC. Spre11gstoffw.-Nltrocell~~lose,
33: 41. 1938.