?

Western State Sectioflhe Combustion Institute 1997 Spring Meeting

April 14 & IS,Livermore, California
SAND97-0186C

Semiconductor Bridge, SCB, Ignition Studies

of AI/CuO Thermitet

Robert W. Bickes, Jr. and David E. Wackerbarth

Sandia National Laboratories
Albuquerque, NM 87185-1453 USA

Jonathan H.Mohler
Energetic Materials Associates, Irtc.
Vero Beach, FL 32968 USA

ABSTRACT

We briefly summarize semiconductor bridge operation and review our ignition studies of
N C u O thermite as a function of the capacitor discharge unit (CDTJ) firing set capacitance,
charge holder material and morphology of the CuO. Ignition thresholds were obtained using
a brass charge holder and a non-conducting fiber-glass-epoxy composite material, G10. At -
18 C and a charge voltage of 50V,the capacitance thresholds were 30.1 id?and 2.0 pF’
respectively. We also present new data on electrostatic discharge (ESD)and radio frequency
(RF‘)vulnerability tests.

INTRODUCTION

Ignition difficulties exhibited by many thermite compositions result from their high thermal
conductivity In particular, hot-wire ignition is especially affected because the temperature
rise of the wire is slow and energy transfer from the wire to the thermite is controlled by
conduction. Consequently a large mass of material outside of the critical ignition zone is
unnecessarily heated before achieving ignition, which increases the amount of energy
required from the firing set.

Semiconductor bridges, SCBs, on the other hand, form a plasma discharge in microseconds
which rapidly transfers energy via a convective process to the thermite. This produces a high
power density in the ignition zone with less energy loss to surrounding materials.
Consequently, thermite compositions can be exploited for pyrotechnic igniters without paying
an ignition energy penalty by using an SCB. Through proper choice of charge holder and
component materials, electrical ignition energies that rival more traditional pyrotechnic
igniter compositions can be achieved.

Statistical ignition-data presented in this paper demonstrate these effects and illustrate the
performance of the SCB in contact with thermite compositions. We also found that by
changing from a metal charge holder to a non-metallic charge holder a s i w c a n t reduction
in ignition energy was achieved. We also present data on electrostatic discharge (ESD) and
radio fiequency @I?) vulnerability tests for SCB-thermite devices.

+“his work was supported by the United States Department of Energy under Contract DE-AC04-94AL8500.
Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the
United States Department of Energy.
 ?

SEMICONDUCTOR BRIDGE, SCB

Sandia's semiconductor bridge, SCB, has three forms that were patented in 1987' and
1990.* Devices utilizing the 1987 patent have been incorporated into Sandia systems for
the Department of Energy (DOE), for the Department of Defense (DoD), and for commercial
systems?

The device described in the 1987 patent is shown in Fig. 1. It consists of a small doped
polysilicon (or silicon) volume formed on a silicon (or sapphire) substrate. The length of the
bridge is determined by the spacing of the aluminum lands seen in the figure. The lands
provide a low ohmic contact to the underlying doped layer. Wires ultrasonically bonded to
the lands and the electrical feedthroughs on the explosive header permit a current pulse to
flow from land-to-land through the bridge; the ultrasonic process produces very strong bonds
and is a cost effective procedure. The doped layer is typically 2 pm thick; bridges are
nominally 90 pm long and 270 pm wide. Bridge resistance at ambient conditions is 1 R;
however, the bridge dimensions can be easily altered to produce other resistances or shapes
for specific applications.

A current pulse through the SCB causes it to burst into a bright plasma discharge that
heats the exoergic material pressed against the bridge by a rapid and efficient convective
p r o ~ e s s .Consequently,
~ SCB devices operate at input energies typically less than 5 mJ (and
as low as 30 pJ). But the SCB devices function very quickly producing an explosive output in
less than 60 ps for pyrotechnic devices (where 60 ps is the interval fiom the firing pulse to
the usable explosive output of the device).

Figure 1. Simplified sketch of a

semiconductor bridge (SCB).
The bridge is formed out of the
heavily doped polysilicon layer
enclosed by the dashed lines.
The bridge dimensions are 270
pn wide (w) by 90 pn long (L)
by 2 pm thick (t). Electrical
leads are attached to the
aluminum lands, permitting an
i applied current pulse to flow
from land-to-land through the
N A-A
bridge.

Despite the low energy for ignition, the substrate provides a reliable heat sink for excellent
no-fire levels. In addition, and as described later the
devices are ESD (electrostatic discharge) and RF (radio frequency) tolerant Because the
physics of SCB operation is so very much different than for hot wires, SCB devices have both
low input energy requirements and high no-fire level^.^

EXPERIMENTAL DESIGN

The following is a brief summary of recent experiments to measure the ignition threshold for
thermite devices.'j Type 50B1 SCB die were mounted on standard TO-46 transistor
headers; bridge dimensions were 90 pm long and 270 pn wide. For some of the tests the
charge holders were brass cylinders pressed and glued onto the TO-46 header.' Figure 2
shows a cross section of these devices. The devices were loaded with 184 mg of N C u O

Page -2-
 DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the United
States Government. Neither the United States Government nor any agency thereof, nor
any of their employes, make any warranty?express or implied, or assumes any legal liabili-
ty or responsibility for the accuracy, completeness, or usefulness of any information, appa-
ratus, product, or process disdased, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product, process, or service by
trade name, trademark, manufacturer, or otherwise does not necessarily constitute or
imply its endorsement, recommendation, or favoring by the United States Government or
any agency thereof. The views and opinions of authors expressed herein do not necessar-
ily state or reflect those of the United States Government or any agency thereof.
Portiom of this document may be illegible
in electronic image products Images are
produced from the best available original
document.

\
thermite and pressed to a density of 2.3 glcm3 (45% of TMD). We used spherical, atomized
aluminum,s and 13600 copper oxide which was commercially processed by air oxidation of
fine particle copper.9

Our primary goal was to obtain the threshold for ignition as a function of CDU capacitance
at a charge voltage of 50 V and with the capacitor and the device both at 0°F. We used the
Neyer SENSIT program to determine the capacitor levels and the ASENT program to
analyze the data.”

Figure 2. TO-46transistor base with a brass charge

holder. The internal diameter of the charge holder is 0.150”
(3.8 mm) and the internal length is 0.270”(6.9 mm). The
outside diameter of the charge holder is 0.25” (6.4 mm).

The f i n g set for these studies was a capacitor discharge

unit (CDU)consisting of a capacitor, switched by a n SCR,
into a series circuit with the SCB and a current viewing
resistor, CVR (see Fig. 3). The capacitor was external to
the firing set circuit box and was connected to the circuitry
with banana plugs; this permitted us to easily change the
capacitor. We used ceramic capacitors from AVX
Corporationll for this study; the capacitances ranged from

WOW
Rl sur
1 1 I <

Figure 3. CDU firing set. Cl is the discharge capacitor and is 5 pF in this illustration.
Current through the bridge is monitored with the current viewing resistor (CVR). A 5 V , 10 p
trigger fires the switch which can be an SCR as shown or an FET.

In addition to measuring the current, the voltage across the SCB and the CVR was also
measured. The waveforms were captured with a Tektronix TDS 640 Digitizer and analyzed
with a DEC LSI 11/73 computer system. For each test we calculate the energy delivered by
the firing set to the SCB. Function times were obtained by a photodiode that viewed the
flash when the unit fired; the times reported are the time interval from the trigger signal to
the iiring set to the light flash.

Page -.3-
First Test Series

Table I. summarizes our first test series. For this 15 unit series the mean all-fire
capacitance was 30.1 pF & 0.2 pF; at a 95%confidence level the 0.999 upper limit was 32.6
pF. Note the small sigma value characteristic of SCB devices

TABLE I: Ignition Data for Brass Charge Holder,

Capacitor Voltage of 50 V, Units at 0°F

Capacitor Function Capacitor Function Capacitor Function

0 Time @) 0 Time (p) 0 Time u s )

11.67 No Go 15.36 No Go 19.07 No Go

23.50 No Go 39.2 93 1 33.0 2328
29.3 No Go 31.9 1352 30.1 No Go
31.9 3514 30.1 2600 29.3 0
30.2 742 29.3 No Go 30.0 No Go

Second Test Series

This test series mirrored the first with the exception that the charge holder was G10, which
is a non-electrically conductive, fiber-glass-epoxycomposite (see Fig. 4). Again our goal was
to determine the threshold capacitance of the CDU firing set at 50 V on the charge capacitor.
These tests were all carried out at O’F. The charge holder was loaded with 200 mg of
AVCuO thermite again pressed to a density of 2.3 g/cm3. The results are summarized in
Table 11. Using the SENSIT and ASENT programs we determine the mean all-fire
capacitance to be 2.08 pF f 0.45 pF; at a 95%confidence level the 0.999 upper limit was
6.30 pF.

Figure 4. G I 0 charge holder. l%e internal diameter is 0.15” (3.8 mm) and the length is 0.36”
(9.1 m m).
 Table II: Ignition Data for a G10 Charge Holder,
Capacitor Voltage of 50 V, and Units at 0°F

Capacitor Function Capacitor Function Capacitor Function

W) Time (ps) w9 Time @) 0 Time @)

15.63 786 11.61 565 7.26 78 1

1 No Go 4.83 672 3.34 567
2.11 No Go 3.34 626 1 No Go
2.03 65 1

Next we carried out a 10 unit SENSIT study to determine the voltage threshold for a 3.34
*
p F CDU for units again fired at 0°F. We obtained a mean voltage of 36.76 V 0.8 V at a
95% confidence level the 0.999 upper limit was 4.22 V (see Table 111.)

Table 111: Ignition Data for a G10 Charge Holder,

3.34 pF CDU Capacitor, and Units at 0°F

Charge Function Charge Function Charge Function

Voltage (v) Time @) Voltage (v) Time @) Voltage (V) Time @)

32.5 No Go 36.2 No Go 45 846

40.6 65 1 38.4 1257 35.1 No Go
37.3 No Go 37.8 2513 38.5 843
36.7 1389

Third Test Series

The goal of this test series was to measure the ignition threshold at O'F but with UP 13600
CuOUversus the 13600 CuO used for the previous experiments. (The 13600 and UP 13600
are company designations that refer to different processes for preparing the CuO; see
references for details.) The aluminum was the same material used previously. The charge
holder was the G10 material and was loaded with 200 mg of the new thermite formulation
pressed to the same density (2.3 g/cm3) as in the previous series,. We b e d 10 units
according to the Neyer program and determined the mean all-fire capacitance to be 2.07 JJ.F;
the data were degenerate (no crossover) thus error and confidence limits were not obtained.
The data are summarized in Table IV.

Table IV: Ignition Data for UP 13600 CuO,

a G10 Charge Holder and a 50 V CDU

Capacitor Function Capacitor Function Capacitor Function

W) Time (ps) W) Time (ps) W) Time @)

4.83 945 3.34 977 2.01 1273

1.17 No Go 1.905 No Go 2.63 1148
1.53 No Go 2.07 No Go 2.27 1141
1.84 No Go

Page -5-
We then carried out a 10 unit SENSIT study to determine the voltage threshold for a 3.34
pF CDU for units fired at O'F. We obtained a mean voltage of 38.4 V; again the data were
degenerate. The data are summarized in Table V.

Table V: Ignition Data for UP 13600 CuO,

a G10 Charge Holder, and a 3.34 p.F CDU Capacitor

Charge Function Charge Function Charge Function

Voltage (V) Time 019) Voltage (V) Time 019) Voltage cv> Time (ps)

32.5 No Go 36.2 No Go 39.9 1532

38.10 No Go 39.0 1422 39.0 No Go
38.6 1590 37.6 No Go 39 2146
37.8 No Go

The data for all three test series is summarized in Table VI. The function times and SCB
energies reported are the average times for the shots that fired.

TABLE VI: SUMMARY

Shot CUO Cap. Cap. charge SCB Function
Nos. Size Volts Holder Energy Time
0 0 (ma ols)
792-806 13600 30.1 50 Brass 28. 2019
941-950 13600 2.1 50 G10 3.1 664
1114-1123 UP 13600 2.1 50 G10 2.3 1097

951-960 13600 3.3 37 G10 1.8 1250

1124-1133 UP 13600 3.3 38 G10 2.2 1673

963 13600 8.7 25 G10 1.7 84 1

1134 UP 13600 8.7 25 G10 1.8 No Go

962 13,600 19.3 20 G10 1.6 784

1135 UP 13600 19.3 20 G10 1.7 1343

961 13600 19.3 17 G10 1.6 No Go

1136 UP 13600 19.3 17 G10 1.5 No Go

940 13600 19.3 50 G10 3.9 629

1111 UP 13600 19.3 50 G10 4.5 957
1110 UP 13600 3.35 50 G10 3.4 1262

W: Shots 940,1110 and 1111 used room temperature units;all other tests were at 0°F.
Fourth Test Series

Based on the previous results, Energetic Materials Associates, Inc. designed and built
components for their commercial applications. Units were sent to Sandia for testing in ESD
and R F environments.

The Sandia Standard Man ESD test places the unit under test in series with a 500 R
resistor and a 600 pF capacitor. The capacitor is charged to 20 kV and then discharged
through the unit and 500 R resistor using a fast rise-time switch. We tested units at 20, 15
and 10 kV.The applied pulses were pin-to-pin, that is the current pulse was through the
SCB as opposed to a pin-to-case test. We also submitted 2 units to the standard DoD ESD
test. For the DoD tests the units are in series with a 5000 R resistor and a 500 pF
capacitor charged to 25 kV.The results are summarized in Table VII. None of the units
fired when subjected to the ESD pulses. However the Standard Man test at 20 and 15 kV
did damage the bridge as indicated by the change in resistance.

All of the units were fired using a 10 pF firing set charged to 45 V. All but one of the units
functioned properly. However, the current waveform for the no-fire indicated that the unit
was not loaded properly; the unit will be dissected and examined to determine the cause of
the no fire.

TABLE VII: ESD TESTS

Voltage Resistance Resistance Function

before test after test Go/NoGo
&W (n) (0)
G10 20 1.018 9.54 Go
G10 20 1.001 10.91 Go
G 10 15 1.008 2.089 Go
G10 10 1.007 1.010 Go
Steel 20 1.012 7.396 Go
Steel 20 1.049 4.621 Go
Steel 15 1.015 1.642 Go
Steel 10 1.026 1.016 NoGo

Two of the steel units were injected with 1 Watt of RF power at 10 MHz input frequency for
5 minutes. The resistances of the units increased slightly after the RF injection but none of
the units functioned (see Table VIII). A third units was subjected to 2, 3, 4 and 5 watt
injections, each 5 minutes long; again the unit did not fire.

The two units tested a t 1 Watt were fired using the 10 pF, 45 V firing set and functioned
properly (Table VIII). The third unit was not tested but will be opened and examined to
determine if the RF injection caused any damage to the bridge or the powder interface.

Page -7-
 TABLE VIII. RF TESTS

TYPe RF Power Resistance Resistance Function

0 before Test after Test Go/NoGo

Steel
(a
1.205
(Q)
1.255 Go
Steel 1.150 1.198 Go
Steel 1.322 1.308
(repeat) 1.308 1.378
(repeat) 1.378 1.346
(repeat) 1.346 1.335

SUMMARY

The mean all-fwe thresholds for these devices were signrficantly influenced by the type of the
holder. The all-fire capacitance for the G10 units (at 50 V) was approximate one-tenth that
of the capacitance for the brass units. The reduction in threshold capacitance may be due to
the thermal losses at the metal charge holder or, based on comparisons of the current and
voltage waveforms, may also be due to the effect on the plasma of the ground plane provided
by the brass cylinder. Operation with the G10 header at 50 V and with a 3 to 6 pF
capacitor provides for a large margin for reliable device function. Powder morphology affects
function time but not ignition sensitivity. The 13600 material produced an output
approximately 1.5 times faster than the UP 13600 CuO. All of the units survived ESD and
RF injection and all but one of the units functioned properly when fired using a 10 pF 45 V
firing set.

ACKNOWLEDGMENTS

This work was sponsored by Sandia‘s Technical Assistance Small Business Initiative. Our
thanks to SCB Technologies Inc. for supplying the G10 charge holders.

For additional information about SCB technology, please contact Bob Bickes, Sandia
National Laboratories, (505)844-0423 or rwbicke@sandia.gov.

REFERENCES

* R. W. Bickes, Jr., A. C. Schwarz, “Semiconductor Bridge (SCB) Initiator,” U.S. Patent

4,708,060 (Nov. 1987).

D. A. Benson, R. W. Bickes, Jr., R. W. Blewer, “A Tungsten Bridge for the Low Energy
Ignition of Explosive and Energetic Materials,” US.Patent 4,976,200 (Dec. 1990).

R. W. Bickes, Jr. M. C. Grubelich, S. M. Harris, W. W. Tarbell, “Semiconductor Bridge,

SCB, Ignition of Energetic Materials,” Western State SectionPThe Combustion Institute 1997
Spring Meeting, April 14 & 15, Livermore, California, SAND97-0188C.

D. A. Benson, M. E. Larsen, A. M. b n l u n d , W. M. Trott, and R. W.Bickes, Jr.,

“Semiconductor Bridge (SCB): A Plasma Generator for the Ignition of Explosives,” Journ.
Appl. Phys. a@), 1622-1632, (Sept. 1987).

Page -8-
 R. W. Bickes, Jr., S. L. Schlobohm and D. W. Ewick, “Semiconductor bridge (SCB) Studies:
I. Comparison of SCB and Hot-wire Pyrotechnic Actuators,” 13* International Pyrotechnics
Seminar, p. 69, Grand Junction, Colorado (July 1988)

R. W. Bickes, Jr., M C. Grubelich, S. M. Harris J. A. Merson, J.H. Weinlein, “An Overview

of Semiconductor Bridge, SCB, Applications at Sandia National Laboratories”, 3lst
WASME/SAEVASEE Joint Propulsion Conference and Exhbit, American Institute of
Aeronautics and Astronautics, May 1995.

These assemblies were purchased from SCB Technologies, Inc., Albuquerque, NM, Part
number 50BllTlBG.

Valimet Inc., Stockton, California, H-3 material with a mean diameter of 3.8 pm.

American Chemet Corp., Deerfield, Illinois, type 13600 copper oxide is prepared by air
oxidation of fine copper particles.

lo B. T. Neyer, “More Efficient Sensitivity Testing,” EG&G Mound Applied Technologies,

MLM-3609, (October 20, 1989).

*I AVX Corporation, X7R Type, Myrtle Beach, South Carolina.

IzAmerican Chemet Corp., Deerfield, Illinois, type UP 13600 copper oxide is processed using
a proprietary procedure to obtain CuO from precipitated CUCOS.

Page -9-