PHYSICAL REVIEW E 86, 066409 (2012)

Exploding-wire experiments and theory for metal conductivity evaluation in the

sub-eV regime

J. Stephens and A. Neuber

Center for Pulsed Power and Power Electronics, Texas Tech University, Lubbock, Texas 79409, USA
(Received 28 September 2012; published 18 December 2012)
Copper and silver wires are subjected to pulsed high current densities producing high density metal plasma
in the sub-eV regime with atmospheric air as a background gas. Numerical simulation via application of the
one-dimensional magnetohydrodynamic partial differential equations solved simultaneously with the constraining
circuit equations is presented. The simulations require accurate knowledge of the material equation of state
(EOS) and transport properties; the LANL SESAME database is applied for the EOS in all cases. Two electrical
conductivity models are applied. First, the Lee-More-Desjarlais (LMD) and its modification, the quantum LMD
(QLMD) conductivity, which have been well proven at higher temperatures, are applied. Simulations with the
LMD and QLMD data indicate that the conductivity data as well as the MHD methodology are accurate in the
sub-eV regime of interest. A less computationally involved, empirical conductivity model is applied in the same
regime to explore its temperature-density range of applicability compared to the more sophisticated model.

DOI: 10.1103/PhysRevE.86.066409 PACS number(s): 52.80.Qj, 52.27.Gr, 52.30.Cv, 52.50.Nr

I. INTRODUCTION metal plasma formation within a few microseconds. In each

experiment, the exploding wire has a 127-μm diameter and
Current induced electrical explosions of conductors have
18-cm length.
become a common mechanism for both producing experimen-
The circuit shown in Fig. 1 is modeled as a series resistor
tal electrical conductivity data [1–8] and verifying existing
inductor capacitor (RLC) circuit. Experimentally, a short
conductivity models [9–13] for strongly coupled metal plasma.
circuit was compared with the numerical simulation to verify
Similar to the methodologies of Refs. [9–13], this paper
the accuracy of the circuit model.
compares voltage and current waveforms obtained from
Parasitic inductance in the experimental setup has to be
exploding-wire (EW) experiments, with magnetohydrody-
considered when the resistive voltage drop across the EW
namic (MHD) simulations, to assess the accuracy of the
is desired. As a result, the following equation is used to
equation of state (EOS) and transport models applied within
derive the resistive voltage drop across the EW from the
the MHD simulation.
charging voltage and pulsed current, with C = 1.85 μF and
In practice, EWs are widely used as high current opening
L = 2.7 μH.
switches [14–20] owing to the sharp decrease in conductivity
experienced between vaporization and the critical point. Silver 
1 t   di(t)
and copper are generally accepted as the optimum switching VR (t) = V0 − i(t )dt − L . (1)
C 0 dt
materials [19,20] and are the materials under investigation
here. The experiments presented exhibit a specific emphasis
on the sub-eV (<11.5 kK) switch type current interruption
process, a region which has received little attention compared III. THEORETICAL MODEL
to its higher temperature counterpart. It is known that EWs demonstrate significant stratification
It is a common practice to submerge the EW in a semi- along the axis of the wire [14,21–23]. Abramova et al. present
incompressible media, such as water, for studies such as these. that the time for MHD instability development can be approx-
Under these conditions, the wire expansion is restricted, thus imated by τMHD ≈ 2(ρ/μ0 )1/2 /jmax , which is on the order of
limiting hydroexpansive cooling of the plasma column and 0.5–1 μs for the experiments discussed here [21]. In these
allowing for the production of extremely high temperatures studies, the wire explosion is a multiple-microsecond process;
(e.g., >90 kK in Ref. [11]). Since this study is primarily aimed therefore, it is expected that the instabilities will have sufficient
at sub-eV temperatures, the desired parameter range is simply time to develop. However, the wires generally exhibit an
achieved in air at atmospheric pressure. approximate, or average, cylindrical behavior. Consequently,
one-dimensional numerical methods have been widely applied
II. EXPERIMENTAL SETUP with a high level of success [8–16,24–26]. Although alterna-
tives to MHD have been successfully applied to model explod-
The experimental setup given in Fig. 1 is used to drive ing conductors [14,27], MHD is the method of choice here.
the EWs described in these experiments. The capacitor shown The one-dimensional, radially directed MHD equations in
in Fig. 1 is initially charged to a controlled voltage V0 . At a Lagrangian coordinates are given as [28]
controlled time, the thyratron switch is closed. The switch
used requires eight parallel, short (∼5 ns) RG213 coaxial ∂ρ ∂ρ ∂v ρv
+v +ρ + = 0, (2a)
transmission lines to uniformly distribute the current density ∂t ∂r ∂r r
inside of the thyratron. Upon closing of the switch, several ∂v ∂v ∂p
kiloamps of current are passed through the EW, inducing dense ρ + ρv = −jz Bϕ − , (2b)
∂t ∂r ∂r

1539-3755/2012/86(6)/066409(5) 066409-1 ©2012 American Physical Society

J. STEPHENS AND A. NEUBER PHYSICAL REVIEW E 86, 066409 (2012)

FIG. 1. Exploding-wire experimental setup.

∂u ∂u ∂v pv j2
ρ + ρv = −p − + z
∂t ∂r ∂r r σ
 
1 ∂ ∂T
+ rκ , (2c)
r ∂r ∂r
 
∂Bϕ ∂ 1 ∂  
= rBϕ , (2d)
∂t ∂r μ0 σr ∂r
1 ∂(rBϕ )
jz = . (2e)
μ0 r ∂r
Here ρ, v, and u are the mass density, velocity of expansion,
and specific internal energy, respectively. jz is the axial current
density, Ez is the axial electric field, where jz = σ Ez , and
Bϕ is the azimuthal magnetic field, and μ0 is the magnetic FIG. 2. (Color online) MHD simulated waveforms. V0 = 15 kV,
permeability. The temperature of the wire, T (ρ,u) is taken to and the copper wire is 18 cm in length with a 127-μm diameter. MHD
be a function of density and internal energy, as determined by simulation with QLMD data is shown by the solid red line, and MHD
the SESAME database [29]. p(ρ,T) and σ (ρ,T) are the material simulation with HK data is shown by the dashed blue line.
pressure and electrical conductivity, both as a function of
temperature and density. The internal pressure of the material is
determined from interpolations of the SESAME data [29]. Due to
the time scales of interest, very little thermal conduction is ob- magnetic field exhibits a steep radial gradient. Knoepfel indi-
served, and it is sufficient to roughly approximate the thermal cates that the magnetic diffusion time can be approximated by
conductivity κ(ρT) by the Wiedemann-Franz law [30]. Note τd ≈ r 2 μ0 σ ≈ 320 ns [33]. The simulated magnetic diffusion
that such treatment of the thermal conductivity assumes that time is generally on the order of 100–200 ns. This slightly
the predominant thermal conduction mechanism is through shorter time scale is the result of Joule heating of the EW,
degenerate transport processes. In similar MHD experiments, thus allowing for more rapid diffusion of the magnetic field.
it was indicated that the EOS of the background gas can be For the time scales of interest here, magnetic diffusion plays
roughly approximated, without jeopardizing the accuracy of a fairly insignificant role. Nonetheless, magnetic diffusion is
the simulation [13]. For these simulations, the EOS of the accounted for in the simulation.
background gas is determined using ideal gas approximations, Equations (2a)–(2e) are solved explicitly using multipoint
and the electrical conductivity is assumed to be zero. It was finite difference techniques [34] with a Lagrangian coordinate
found that a better fit between experiments and MHD simula- system [28,33]. The MHD equations are solved simultaneously
tions was achieved by decreasing the simulated wire length by with the constraining circuit equations to derive the voltage and
∼10%. It is suspected that this is the result of plasma formation current waveforms, along with the physical parameters of the
at the contact electrodes, thus shunting any EW behavior in EW throughout the dense plasma formation (cf. Fig. 2).
these regions, similar to what is observed in Refs. [31,32]. Two different conductivity models are applied for both
Equations (2a)–(2c) define the dynamics of the EW, from silver and copper. For silver, conductivity from the Lee-More-
which the temperature and density can be derived. With Desjarlais (LMD) [35,36] algorithm is applied. Copper uses
these parameters, the remaining material properties can be the quantum LMD (QLMD) data, which involves fitting the
determined. Equation (2d) defines the diffusion of the magnetic LMD algorithm to the quantum molecular dynamics (QMD)
field into the EW. Note that the given equation neglects the simulation methods of Ref. [37]. The LMD and QLMD data
effects of the EW expanding into the magnetic field. This is a have been verified to be very accurate in high temperature EW
reasonable assumption given that very little expansion of the experiments; however, very little testing has been conducted in
EW is experienced on the initial time scales during which the the sub-eV regime. The second conductivity model applied is

066409-2
EXPLODING-WIRE EXPERIMENTS AND THEORY FOR . . . PHYSICAL REVIEW E 86, 066409 (2012)

based on the empirical equation given by Knoepfel (HK) [33]:

 α
σ0 ρ
σH K (ρ,T ) = (3)
1 + β(T − T0 ) ρ0
Here T0 is the room temperature, 298 K, ρ0 is the mass density
at room temperature, and σ0 is the electrical conductivity
at room temperature. The parameters α and β are the
mass density and temperature coefficients. Experiments and
comparison with the LMD and the QLMD data indicate α =
2.4 and β = 0.002 for copper and α = 2.6 and β = 0.009 for
silver provide the best performance from the HK conductivity.
Note that the values given here deviate slightly from those
presented by Knoepfel [33]. It is suspected that this is due to the
fact that the values given here are specifically optimized to our
limited temperature-density range, whereas Knoepfel’s values
are intended to cover a wider range than is discussed here.
As the LMD and QLMD conductivities are based on
inherently more complex models, they have a significantly
wider range of validity. Alternatively, a consequence of the FIG. 3. (Color online) Experimental and simulated voltage and
HK conductivity’s simplicity is the limited parameter range current waveforms. V0 = 7.5 kV, and the copper wire is 18 cm in
for which it is accurate. The focus of this paper is to confirm length with a 127-μm diameter. Experiment is shown by the solid red
the LMD and QLMD conductivity in the sub-eV regime and line, MHD simulation with QLMD data is shown by the dashed blue
similarly explore the validity of the HK conductivity. line, and MHD simulation with HK data is shown by the dash-dotted
It was shown by Lindemuth et al. that if the exploding green line.
conductor expansion is not sufficiently limited, the rapidly
growing cross section will ultimately result in decreases voltage across the EW [cf. VR in Eq. (1)] drops sharply.
in resistance [27]. However, with the HK conductivity, the At this time, the current is also sufficiently low that the
resistance begins to diverge to infinity at lower density. It capacitor is only slowly losing charge through current flow, and
is therefore concluded that the HK equation fails at lower the voltage across the EW becomes approximately constant.
density (significant error below 1 gm/cm3 ) and is incapable of Figure 2 also demonstrates that by exploding the wires in
properly modeling cross-sectional growth induced resistance air at atmospheric pressure, expansive cooling of the EW is
decreases in EW experiments. Additionally, it is known that critical to the fundamentals of the experiments shown here, as
for certain conditions, the plasma conductivity will increase is demonstrated by the decaying EW temperature after 2 μs.
with increasing temperature [36–39], yet another phenomenon
not captured by the HK conductivity.
IV. RESULTS
The results of a MHD simulation for typical EW parameters
are given in Fig. 2 for both QLMD conductivity and HK con- Experimental and simulated voltage and current wave-
ductivity. The “average conductivity” in Fig. 2 was calculated forms, as well as the current to voltage ratio for copper wire,
using σAv (t) = I (t)Z0 /[VR (t)Aw (t)], where Z0 is the wire with an initial capacitor voltage of 7.5 kV are depicted in
length and Aw (t) is the time-dependent wire cross-sectional Fig. 3. For an initial voltage of 7.5 kV, it is observed that
area. As shown, the QLMD and HK conductivity models the current excitation provided by the system is insufficient to
demonstrate the same approximate behavior. The current drive the EW through the steep resistive transition. As shown
initially follows an underdamped RLC type discharge. During in Fig. 3, MHD simulation with the QLMD conductivity is able
this time, the conductivity of the wire is still very high, and to accurately capture the behavior of the EW. Similarly, MHD
Joule heating is minimized, as is demonstrated by the lower simulation with the HK conductivity is able to come close to
temperature for times less than 1 μs. As the temperature of the experimentally observed behavior but to a lower degree of
the wire increases, conductivity decreases, and the rate of accuracy than the QLMD data. Comparison of the I -V ratio
Joule energy deposition rapidly increases, thus quickly driving correlates to a comparison of the time evolution of the electrical
the temperature beyond 6 kK. Note that the vaporization conductivity, given that the wire radius from the MHD simula-
temperature of copper is approximately 2843 K [40], and tions and experiment all roughly agree with one another. As is
the critical temperature is approximately 8390 K [39]. This shown in Fig. 3, the same general behavior of the I -V ratio for
increase in temperature also corresponds to rapid expansion both MHD simulations and experiments indicates that the elec-
of the EW plasma column. This sharp launch to higher trical conductivity is reasonably represented by either model.
temperature and lower density results in a substantial decrease Figure 4 depicts an additional experiment with a copper EW
in electrical conductivity. As a result, the resistance of the and a charging voltage of 20 kV. With this excitation, the EW
EW increases to several orders of magnitude times its initial undergoes the rapid resistance increase similar to what would
value. The corresponding increase in EW resistance forces be observed in an EW based opening switch. MHD simulations
a sharp interruption in EW current and large EW voltage, with the QLMD conductivity are able to predict this behavior
corresponding to the induced di/dt. At later times, once reasonably well; however, the time of current interruption
the current is completely interrupted, i.e., di/dt ∼ 0, the is slightly underestimated. With the HK conductivity, the

066409-3
J. STEPHENS AND A. NEUBER PHYSICAL REVIEW E 86, 066409 (2012)

data. Similarly, the HK based MHD simulation is also able to

follow the experimental data, to a reasonable degree. Again,
in evaluating the I -V ratio it is observed that the LMD
based MHD simulations capture the experimentally observed
conductivity evolution. However, again, the HK based MHD
simulation overpredicts the conductivity by roughly an order
of magnitude.
It should be noted that the overall quantitative error in the
calculations is presently undetermined, primarily due to the
various potential limitations of the techniques presented here.
Nevertheless, some study of uncertainty in wire parameters
exists in the open literature (refer to Drake et al. [41]): Using
MHD and statistical variation of wire radius, length, and
resistance, the effects of uncertainty in the wire parameters on
the resulting simulated voltage and current waveforms, com-
pared to experimental data, are comprehensively discussed.
It is possible that similar uncertainty in the wire parameters
for the experiments discussed here contributes to the error
FIG. 4. (Color online) Same as Fig. 3, with V0 = 20 kV. in these MHD simulations. Additionally, the application of
the arbitrary phase, temperature-density controlled model also
peak current is overpredicted, and again, the interruption has limitations. Applying such models, especially in the two
time is underestimated. Nonetheless, the overall dynamics phase liquid-vapor region below the critical point, has been
of the explosion are captured with the HK conductivity, questioned before.
despite the simplicity of the model. The I -V ratio given
in Fig. 4 indicate that, although the time at which the V. CONCLUSION
conductivity transition occurs is slightly underestimated, the
MHD simulation with the QLMD conductivity accurately Ultimately, several experiments were run at charging
captures the conductivity evolution experimentally observed. voltages of V0 = 7.5, 10, 15, and 20 kV for both copper
Simulations with the HK data indicate that the conductivity is and silver. In all cases, the LMD and QLMD based MHD
overestimated by as much as an order of magnitude in the late simulations demonstrated approximately the same level of
stages of the experiment. Regardless, as stated, the voltage and accuracy summarized in Figs. 3–5. In general, the accuracy of
current waveforms are fairly well replicated when applying the the methodology described here is strongly dependent upon the
HK conductivity. accuracy of the SESAME data, validity of the one-dimensional
Similar observations have been made with experimentation treatment of the EW, and numerical techniques applied.
and simulation of silver EWs. Figure 5 shows the voltage and Regardless, given the reasonable fit between experimental and
current waveforms for an initial voltage of 15 kV. Again, Joule simulated waveforms, it is concluded that the LMD and QLMD
heating of the EW drives the material to the point where a sharp data are suitable for modeling in the sub-eV regime.
drop in current is experienced. As expected the LMD based Expectedly, the HK dependent MHD simulations demon-
MHD simulation is able to closely follow the experimental strated a few issues. In general, the HK conductivity gives
higher conductivity than experimentally observed data, prior
to the current interruption process (i.e., lower predicted
EW voltages and higher currents are simulated prior to
current interruption). Additionally, simulations with the HK
conductivity have a tendency to predict current interruption
where such behavior is not experimentally observed. However,
when current interruption is observed, the HK dependent MHD
predicts much less aggressive current interruption in the final
stages of the interruption process. However, implementation of
an analytical approximation is appealing, given its accessibility
and significant decrease in necessary computing power as
compared to interpolations of large data sets. For example,
in these studies, the MHD simulations using the HK data were
completed in less than a minute on a 32 core server workstation,
roughly one-tenth of the time needed for the LMD and
QLMD MHD simulations. It is possible that the inaccuracies
may be considered acceptable in first-order approximations,
for instance, for quick calculation of fuse opening switch
performance in the early design stages of a pulsed power
FIG. 5. (Color online) Same as Fig. 3, with silver wire, LMD system, again, given the simplicity and low computational
model, V0 = 15 kV. demand associated with the HK data.

066409-4
EXPLODING-WIRE EXPERIMENTS AND THEORY FOR . . . PHYSICAL REVIEW E 86, 066409 (2012)

ACKNOWLEDGMENTS Dr. Desjarlais and Dr. Cochrane of Sandia National Labora-

tories are thanked for providing the LMD and QLMD data
The authors would like to thank Sandia National Labora-
and useful discussions. Los Alamos National Laboratory is
tories, which has, in part, supported this research through a
thanked for supplying the SESAME equation of state database.
fellowship from the National Physical Sciences Consortium.

[1] A. W. DeSilva and H.-J. Kunze, Phys. Rev. E 49, 4448 (1994). [23] G. S. Sarkisov, S. E. Rosenthal, K. R. Cochrane, K. W. Struve,
[2] A. W. DeSilva and J. D. Katsouros, Phys. Rev. E 57, 5945 (1998). C. Deeney, and D. H. McDaniel, Phys. Rev. E 71, 046404
[3] I. Krisch and H.-J. Kunze, Phys. Rev. E 58, 6557 (1998). (2005).
[4] J. F. Benage, W. R. Shanahan, and M. S. Murillo, Phys. Rev. [24] S. I. Tkachenko, K. V. Khishchenko, V. S. Vorob’ev, P. R.
Lett. 83, 2953 (1999). Levashov, I. V. Lomonosov, and V. E. Fortov, High Temp. 39,
[5] V. N. Korobenko, A. D. Rakhel, A. I. Savvatimiski, and V. E. 674 (2001).
Fortov, Phys. Rev. B 71, 014208 (2005); 71, 099902(E) (2005). [25] K. V. Khishchenko, S. I. Tkachenko, P. R. Levashov, I. V.
[6] V. N. Korobenko and A. D. Rakhel, Phys. Rev. B 75, 064208 Lomonosov, and V. S. Vorob’ev, Int. J. Thermophys. 23, 1359
(2007). (2002).
[7] J. Clérouin, P. Noiret, P. Blottiau, V. Recoules, B. Siberchicot, [26] D. Bakulin, V. F. Kuropatenko, and A. V. Luchinskii, Sov. Phys.
P. Renaudin, C. Blancard, G. Faussurier, B. Holst, and C. E. Tech. Phys. 21, 1144 (1976).
Starrett, Phys. Plasmas 19, 082702 (2012). [27] I. R. Lindemuth, J. H. Brownell, A. E. greene, G. H. Nickel,
[8] D. Sheftman, D. Shafer, S. Efimov, and Ya. E. Krasik, Phys. T. A. Oliphant, and D. L. Weiss, J. Appl. Phys. 57, 4447
Plasmas 19, 034501 (2012). (1985).
[9] A. Grinenko, V. T. Gurovich, A. Saypin, S. Efimov, Y. E. Krasik, [28] D. Schnack, Lectures in Magnetohydrodynamics: With an
and V. I. Oreshkin, Phys. Rev. E 72, 066401 (2005). Appendix on Extended MHD (Springer, Heidelberg, 2009).
[10] V. I. Oreshkin, R. B. Baksht, A. Yu. Labetsky, A. G. Rousskikh, [29] Los Alamos National Laboratory, Report No. LA-UR-92-3407,
A. V. Shishlov, P. R. Levashov, K. V. Khishchenko, and I. V. edited by S. P. Lyon and J. D. Johnson, 1992 (unpublished).
Glazyrin, Tech. Phys. 49, 843 (2004). [30] G. Wiedemann and R. Franz, Ann. Phys. 89, 497
[11] D. Sheftman and Ya. Krasik, Phys. Plasmas 17, 112702 (2012). (1853).
[12] D. Sheftman and Ya. E. Krasik, Phys. Plasmas 18, 092704 [31] S. A. Pikuz, T. A. Shelkovenko, D. B. Sinars, J. B. Greenly,
(2011). Y. S. Dimant, and D. A. Hammer, Phys. Rev. Lett. 83, 4313
[13] A. E. Barysevich and S. L. Cherkas, Phys. Plasmas 18, 052703 (1999).
(2011). [32] D. B. Sinars, T. A. Shelkovenko, S. A. Pikuz, M. Hu, V. M.
[14] J. Stephens, A. Neuber, and M. Kristiansen, Phys. Plasmas 19, Romanova, K. M. Chandler, J. B. Greenly, D. A. Hammer, and
032702 (2012). B. R. Kusse, Phys. Plasmas 7, 429 (2000).
[15] J. Stephens and A. Neuber, Phys. Plasmas 19, 060702 (2012). [33] H. Knoepfel, Magnetic Fields: A Comprehensive Theoretical
[16] J. Stephens, W. Mischke, and A. Neuber, IEEE Transac- Treatise for Practical Use (Wiley, New York, 2000).
tions on Plasma Science, Vol. 40 (IEEE, New York, 2012), [34] J. Li and Y. Chen, Computational Partial Differential
pp. 2517–2522. Equations Using MATLAB (Taylor and Francis, Boca Raton, FL,
[17] M. A. Elsayed, A. A. Neuber, J. C. Dickens, J. W. Walter, 2008).
M. Kristiansen, and L. Altigilbers, Rev. Sci. Instrum. 83, 024705 [35] Y. T. Lee and R. M. More, Phys. Fluids 27, 1273
(2012). (1984).
[18] B. M. Novac, I. R. Smith, P. Senior, M. Parker, and G. Louverdis, [36] M. P. Desjarlais, Contrib. Plasma Phys. 41, 267 (2001).
Rev. Sci. Instrum. 81, 054706 (2010). [37] M. P. Desjarlais, J. D. Kress, and L. A. Collins, Phys. Rev. E 66,
[19] G. A. Mesyats, Pulsed Power (Springer, Berlin, 2004). 025401 (2002).
[20] D. R. McCauley, D. W. Belt, J. J. Mankowski, J. C. Dickens, [38] S. Kuhlbrodt and R. Redmer, Phys. Rev. E 62, 7191 (2000).
A. A. Neuber, and M. Kristiansen, IEEE Trans. Plasma Sci. 36, [39] V. E. Fortov and I. T. Iakubov, The Physics of Non-ideal Plasma
2691 (2008). (World Scientific, Singapore, 2000).
[21] K. B. Abramova, N. A. Zlatin, and B. P. Peregud, Sov. Phys. [40] Explosively Driven Pulsed Power, edited by A. A. Neuber
JETP 42, 1019 (1976). (Springer, Berlin, 2005).
[22] A. G. Rousskikh, V. I. Oreshkin, S. A. Chaikovsky, N. A. [41] R. R. Drake, D. R. Koenig, and J. H. J. Niederhaus
Labetskaya, A. V. Shishlov, I. I. Beillis, and R. B. Baksht, Phys. https://cfwebprod.sandia.gov/cfdocs/CCIM/docs/Exploding
Plasmas 15, 102706 (2008). WireUUR_paper.pdf.

066409-5