1100 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO.

3, JUNE 2005

An Equivalent Circuit Model of a

Plasma Core Inductor
Sudarshan Rao Nelatury, Senior Member, IEEE, Thomas Lee Hemminger, Senior Member, IEEE, and
Matthew N. O. Sadiku, Senior Member, IEEE

Abstract—The impedance of an electrical coil wound over a flu-

orescent tube is found experimentally both when the tube is ener-
gized and de-energized. This paper reports results showing a sig-
nificant change in the resonance characteristics of the coil due to
the influence of the plasma core. The effects are described with the
help of an equivalent circuit consisting of an inductance with a se-
ries resistance and a parallel capacitance. This paper demonstrates
that by energizing the core, the series resistance and shunt capac-
itance in the equivalent circuit increase significantly. The experi-
ment is performed both for alternating current and direct current
excitations. From the observed results, fractional changes in the
resonant characteristics are found and explained using a perturba-
tional viewpoint. This enables us to indirectly estimate the plasma
density profile and model the beam loading effects in high-energy
radio frequency sources. The equivalent circuit model presented Fig. 1. Induction plasma torch.
here can be employed to explain the macroscopic effect of low-tem-
perature plasma on circuit elements within close proximity.
plasma torch, the plasma is produced and sustained by induction
Index Terms—Equivalent circuit, inductive plasma, plasma from a time varying magnetic field generated by an induction
loading, T-8 fluorescent tube. coil, as shown in Fig. 1. The coil design is found to affect both
the temperature field and the flow field and to have a significant
I. INTRODUCTION effect on the reaction kinetics in the torch. The rigorous math-
ematical analysis involving Maxwell’s and heat transfer equa-

F OR almost a century, gas discharge plasmas have been a

subject of serious concern that resulted in numerous con-
sumer and industrial applications [1], [2]. It can be said that the
tions for the electromagnetic fields, and for the plasma flow pa-
rameters can be supplemented by an equivalent circuit model
[4]. It has been stated that numerical simulations have virtually
development of atomic physics at the end of 19th century was no impact on industrial practices in this field [2]. The equivalent
virtually synonymous with the gas discharge physics. During circuit approach is much simpler and can furnish a physical un-
the second half of the 20th century, much attention was paid to- derstanding of the effects of the plasma.
ward thermonuclear fusion and space plasma physics. Reactive, This paper provides an equivalent circuit of a coil having
nonequilibrium plasmas have enjoyed a considerable revival of an ionized plasma core. First, we shall start with the equiva-
interest and attention from technologies concerned with elec- lent circuit of an ordinary air-core coil of known dimensions
tronic materials processing and increasingly from the plasma and number of turns, then modify the same when the core be-
physics community. comes ionized plasma. Close agreement is found between the
There are many important applications of partially ionized impedance measured and that calculated from the equivalent cir-
gases. One of the opportunities in the field of applied plasma cuit. Even though our approach is limited to a particular set of
science is in the development of models that can be used for core parameters, it is a reasonable first step in approximating
large-scale simulations of industrial tools. For some period, em- plasma coils with equivalent circuits. As we are interested in
pirical methods have been implemented to design, develop, and the influence of plasma on the coil, we have taken a simple coil
control plasma-processing devices. With the advent of powerful of copper wire wound on an ordinary fluorescent tube and ex-
computers, several numerical techniques have been proposed to perimentally obtained the impedance results both when the tube
analyze and design plasma technology like inductively coupled is de-energized and energized with alternating current (ac), as
plasma torches for industrial applications [3]. In an induction well as direct current (dc) excitation. The equivalent circuit dis-
cussed here appears quite accurate in representing the effects of
the plasma core.
Manuscript received August 26, 2004; revised January 7, 2005.
S. R. Nelatury and T. L. Hemminger are with the School of Engineering and The rest of the paper is organized as follows. Section II deals
Engineering Technology, Pennsylvania State University, Behrend College, Erie, with the equivalent circuit of an ordinary air core coil and the
PA 16563-1701 USA. modification proposed for a plasma core. In Section III, the ex-
M. N. O. Sadiku is with the College of Engineering, Prairie View A&M Uni-
versity, Prairie View, TX 77446 USA. perimental results are given. Conclusions are provided in Sec-
Digital Object Identifier 10.1109/TPS.2005.848618 tion IV.
0093-3813/$20.00 © 2005 IEEE
NELATURY et al.: AN EQUIVALENT CIRCUIT MODEL OF A PLASMA CORE INDUCTOR 1101

Knowing , , , we can get and from (1) and (5).

Conversely, if we know , , and , we can get and .
With a slight algebraic manipulation of the terms in (1) and (5),
the quantity permits us to write
and as

(6)
Fig. 2. Equivalent circuit of an air-core coil.
(7)

II. EQUIVALENT CIRCUIT

Now if the core of the coil is plasma, the equivalent circuit
The equivalent circuit of a coil in its simplest form consists of in Fig. 2 needs to be modified. The plasma we are using in
an inductance , in series with an ohmic resistance . Since the the present work is not generated by high coil current at radio
physical geometry of the conducting surface of the coil is associ- frequency (RF) frequencies, but it is just the ionized plasma
ated with some stray capacitance in parallel, the circuit shown in of a fluorescent tube with a dc voltage within the manufac-
Fig. 2 can be accepted to represent the overall impedance char- turer’s ratings in one experiment, and it is operated from a 60-Hz
acteristics. power line in a second experiment. The idea of using the fluores-
With the capacitance in parallel, the coil essentially exhibits cent light also has the significance that they are ubiquitous and
a resonance characteristic at frequency are fixed in residential, commercial, and nearly all institutional
buildings without any obstruction for proper functioning. So it
is also of interest to know their effect on a circuit component
such as a coil in close proximity. An instance of their interfer-
(1)
ence on personal communication systems is found in [8].
The standard fluorescent tube used in the USA is called a
When the values of and are very small, the above can be “T-12 tube.” This tube, although the most common, is in the
approximated as process of being replaced by a more energy efficient element
called a “T-8 tube” with a 1–in diameter. They are filled with
(2) both mercury (Hg) at a pressure of 3–10 mtorr and argon (Ar)
gas at a pressure of a few torr. The electron emission from the
cathode causes the excitation of the Hg atoms, which in turn
If the length, radius, number of turns, and the core of the coil are results in the emission of 253.7 nm ultraviolet (UV) radiation.
, , , and air, respectively, its inductance can be approxi- The UV radiation is converted into visible light by phosphor
mated by coating on the interior tube wall. The background Ar gas serves
to enhance the ionization efficiency of Hg and to establish and
(3) sustain the arc discharge in the tube. The interaction of the tube
with the coil can best be described by the permittivity of the gas
where is the permeability of air or free space. Equation (3) is mixture in the tube [9], [10]. The electron density , is found
acceptable for longer and tightly wound coils. Shorter coils have based on , where is the current density (typically
greater flux leakage, and hence, a lower value of inductance [5]. 493.38 for a T-8 tube with a line current of 250 mA) and
If the core of the coil is any other homogenous isotropic mag- is the electron charge . The electron drift ve-
netic substance with a permeability , the formula for induc- locity can be found from the known parameters and the mobility
tance in (3) would change accordingly. Also, if the coil is made [1], however, in this work, we made a reasonable approximation
of a conductor of conductivity , diameter , and number of of 1% of its thermal velocity , which depends on the electron
turns , the resistance is [4] temperature [8]. With a typical gas discharge electron temper-
ature of 2 eV, is calculated to be about 6000 m/s. With these
figures, an electron density cm is obtained.
Indeed, an arbitrary choice of the temperature and drift velocity
(4) makes the density calculations questionable. However, the ob-
servations reported in this paper are useful for understanding
at a frequency in Hz. The presence of in the above the relative changes. The weakly ionized gas in the lamp can be
expression is due to the skin effect. The gauge of the wire deter- considered as a cold plasma, therefore, the dielectric constant
mines the value of (4 in case of AWG #27). At resonance, the is a function of frequency and is specified by the well-known
terminal impedance of the coil is real, say , and is given by expression

(5) (8)
1102 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 3, JUNE 2005

Fig. 3. Relative permittivity of the ionized plasma in a typical T-8 versus Fig. 4. Modified equivalent circuit of the coil with plasma core.
frequency at a plasma current of 250 mA.

various frequencies are compared with those obtained by calcu-

where is the plasma frequency in hertz. This can be written lating them for the equivalent circuit shown in Fig. 4.
in terms of free space permittivity and mass of the electron
as [8]
III. EXPERIMENTAL RESULTS
(9) The details of the experiment are now provided. A coil of
approximately turns made of AWG #27 copper wire
With the estimated electron density cm , is wound on a fluorescent tube (T-8) over a length of 17.8 cm
the plasma frequency of the T-8 is nearly 6.43 GHz and the (about 7 in). The wire has a diameter of 0.36 mm (0.0142 in).
relative permittivity is plotted in Fig. 3. The inductance of the coil is about 640 . For this experi-
Note that for frequencies much greater than the plasma fre- ment, assuming , (3) yields a result that is within 1.4%
quency, the ionized medium acts like free space with a relative of the measured value. This is because the coil is relatively long
permittivity of unity, but for frequencies below , the plasma and tightly wound over the core. The dc resistance of the wire
reflects the fields and acts like an electrical barrier. We wish to is 6.025 . These values were calculated and then confirmed
focus on the effect of the plasma on the impedance character- by measurement with a digital inductance–capacitance–resis-
istics of the coil at frequencies lower than . This can be ac- tance (LCR) meter. The actual stray capacitance is small, so to
complished by modeling the coil and plasma system as a trans- be able to measure observable resonance we added a capaci-
former. This idea dates back to Thomson [11], and even recently, tance of 1.049 nF in parallel to the coil. The capacitance at the
it has been successfully used in [4]. According to this approach, terminals was then measured to be 1.082 nF. The coil is con-
the mutual coupling between the plasma and the coil results in nected to a function generator and the voltage across the coil
an equivalent impedance which, when reflected to the coil cir- and current through the coil (with the help of a 1- series re-
cuit can be attributed as an additional series resistance and sistance) are measured at various frequencies. The magnitude of
a parallel capacitance , as shown in Fig. 4 within the ovals. the impedance versus frequency is plotted both when the tube is
The resistance is a measure of plasma loading on the coil. turned off and when it is turned on at input line voltages of 120 V
The capacitance is characterized as an increase of the par- ac 60 Hz, and also dc after the former is full-wave rectified and
allel stray capacitance because the core of the coil is no longer filtered. The voltage across the tube in these two cases is 57 and
air but a conducting medium. Both of these values are assumed 62 V, respectively. There is a significant drop across the choke
to be proportional to the electron density in the plasma, which in the ac case. The plasma current in the tube for these voltages
actually has a nonuniform profile, but can be approximately is 258 and 163 mA, respectively. The measured impedance is
taken as uniform. and are proportional to the root mean plotted in Fig. 5.
square (rms) current through the plasma [4]. Thus, we model With the core energized, the resonant frequency and terminal
them as resistance at resonance were measured. The total series resis-
tance and shunt capacitance were found using (6) and (7). These
(10) values were used to calculate the impedance of the equivalent
(11) circuit as illustrated in Fig. 5 by the dotted lines. The significant
changes in resonance characteristics can be modeled with the
In Section III, the validity of this postulate is verified experimen- modified values of series resistance and shunt capacitance, as
tally, yielding additional verification of the proposed equivalent predicted in the preceding section. The figure illustrates a very
circuit model. The impedance values of the coil measured at strong fit between the measured values of impedance and the
NELATURY et al.: AN EQUIVALENT CIRCUIT MODEL OF A PLASMA CORE INDUCTOR 1103

Fig. 5. Impedance magnitude as a function of frequency. “x” indicates when Fig. 6. Plasma loading as a function of plasma current in mA.
tube is off, “.” when tube is excited by ac from the power line with a line current
of 258 mA, and “o” when the tube is excited by dc with a line current is 163
mA. Dotted lines represent interpolated values of the impedance magnitude, and 0.9843, respectively. A linear relationship is also predicted
using the equivalent circuit parameters in each case. in [4] which helps substantiate the proposed model. From the
fact that , where , , , are, respectively,
TABLE I the density, electron charge, average drift velocity of the elec-
TERMINAL CHARACTERISTICS OF PLASMA CORE INDUCTOR trons, and cross-sectional area of the plasma tube, we find that
the plasma beam loading parameters are related to the density.
When the experiment is repeated by powering the plasma tube
by ac, we found changes in the impedance of the coil, but the
changes in resonant frequency around 164 kHz are not promi-
nent. Xiao and Mo [7] reported that the resonant frequency
in case of Dielectric Cerenkov resonator filled with a periodic
experimental model in the three cases shown in Fig. 5. The re- plasma grating did not change much but they did observe a sig-
sults are summarized in Table I, where it can be seen that the nificant change in the Q of the cavity. Our experience with ac
plasma has a significant and predictable effect on the resonant has confirmed this, but when the tube is powered by dc, we ob-
frequency, and hence, on the equivalent model of the inductor. served a downshift in resonant frequency as reported by Hafizi
To verify the relationship in (10) and (11), the experiment is and Gold [6].
repeated for different values of dc current through the plasma
tube, varying the current between 50–300 mA dc. With the help
IV. PERTURBATIONAL INTERPRETATION
of a variac, the input line voltage is adjusted and then fullwave
rectified and filtered before it is fed to the tube. The resonant fre- A further step can be taken to interpret the aforementioned
quency and terminal resistance at resonance have been noted for observations using a perturbational view point. Perturbational
these current values. Equations (6) and (7) were used to estimate methods are useful for calculating changes in some quantity due
the net values of series resistance and shunt capacitance. Sub- to small changes in the problem. One might estimate possible
tracting the ohmic resistance and the original shunt capacitance changes in the resonance and quality factor of a general cavity
from these numbers, the differential values of and were by computing the fractional change in resonant frequency due
estimated and plotted in Fig. 6. The linear relationship assumed to perturbed values of the electric and magnetic field quantities.
in the (10) and (11), is found by regression on the observations Let the variables , and , represent the electric field,
as magnetic field and resonant frequency of the original cavity
and , and , represent the same quantities in the perturbed
(12) cavity. The effect on the resonant frequency can be represented
by [12]
(13)

where is the estimated plasma loading resistance in ohms,

is the estimated capacitance in picofarad, and is the (14)
plasma current in milliamperes. It should be noted that (12) and
(13) are valid for operating range of only. The y-intercepts where the left-hand side of this quantity is fractional change in
arise from regression and only have numerical significance. The the resonant frequency . Suppose we ignore
correlation coefficients in (12) and (13) are found to be 0.8660 the changes in permeability of the medium and further assume
1104 IEEE TRANSACTIONS ON PLASMA SCIENCE, VOL. 33, NO. 3, JUNE 2005

Fig. 7. (a) Fractional change in the resonant frequency versus that in the capacitance. (b) Calculated plasma density. Note that there is a down shift in the resonant
frequency with increasing n due to beam loading [6].

that the plasma column is a homogenous medium with permit- density is found to bring about a downshift in the resonant fre-
tivity being spatially uniform. The fractional change in the per- quency in case of dc excitation. The overall effect is an added se-
mittivity can then be related to the fractional change ries resistance and shunt capacitance proportional to the plasma
in the resonant frequency as current. This representation permits us to employ equations (12)
and (13) to include circuit elements with a relatively simple cur-
(15) rent measurement of the core. It is very likely this concept can be
where extended to similar plasma devices, and even at other frequen-
cies, as well, wherein resonant circuits are realized as cavities.
(16) Our approach provides an experimental method of evaluating
plasma parameters indirectly just as how one might arrive at
is a constant that represents the negative slope of the straight the same conclusions by employing Maxwell’s equations,
line plot represented by (15). We used to indicate heat transfer theory, and science of plasma physics. This
the intrinsic impedance of the medium. The fractional change model should provide a useful approximation of the effects
in the resonant frequency versus that in the shunt capacitance in of low-temperature plasma on coil inductance and provide a
the present lumped model is plotted in Fig. 7(a). Assuming an starting point for further work on the effects of plasma tubes on
electron temperature of about 2 eV, which results in average ve- surrounding circuit elements. Plasma core modeling is useful in
locity of 6000 m/s, we can calculate the plasma density, which the analysis and design of inductively coupled plasma torches,
is plotted in Fig. 7(b). Note that there is a downshift in reso- and in understanding the effects of fluorescent lighting on
nant frequency with increasing due to beam loading. Strictly electromagnetically sensitive technology. It can be employed
speaking, even the electron temperature would also change, but in certain areas of plasma science, for example, semiconductor
that is not considered in this work. manufacturing, and design of high energy RF sources like the
magnicon to understand the effects of beam loading.
V. CONCLUSION
ACKNOWLEDGMENT
This paper proposes an equivalent circuit for a plasma core
inductor. The model was developed by observing the resonant The authors are grateful to Dr. S. Gitomer for his valuable
frequency and impedance characteristics of the coil using a T-8 suggestions. The authors also acknowledge the helpful com-
fluorescent tube as the plasma core. An increase in the plasma ments of the reviewers.
NELATURY et al.: AN EQUIVALENT CIRCUIT MODEL OF A PLASMA CORE INDUCTOR 1105

REFERENCES Thomas Lee Hemminger (S’87–M’92–SM’05) re-

ceived the B.S. degree from Cleveland State Univer-
[1] M. A. Liberman and A. J. Lichtenberg, Principles of Plasma Discharges
sity, Cleveland, OH, in 1985, the M.S. degree from
and Materials Processing. New York: Wiley, 1994. Clarkson University, Potsdam, NY, in 1988, and the
[2] D. B. Graves, “Plasma processing,” IEEE Trans. Plasma Sci., vol. 22,
Ph.D. degree from Case Western Reserve University,
no. 1, pp. 31–42, Feb. 1994. Cleveland, in 1992, all in electrical engineering.
[3] J. Fouladgar and A. Chentouf, “The calculation of the impedance of an He is currently an Associate Professor at the
induction plasma installation by a hybrid finite element method,” IEEE
Pennsylvania State University, Behrend College,
Trans. Magn., vol. 29, no. 6, pp. 2479–2481, Nov. 1993. Erie. His research interests include multidimensional
[4] J. T. Gudmundsson and M. A. Lieberman, “Magnetic induction and
signal processing, spectral analysis, imaging sys-
plasma impedance in a cylindrical inductive discharge,” Plasma Sources tems, electromagnetic properties of materials, and
Sci. Technol., vol. 6, no. 4, pp. 540–550, Nov. 1997.
neural networks. He presently devotes much of his activities to the development
[5] F. W. Grover, Inductance Calculations. New York: Dover, 1946. of routing algorithms in communications networks and to the application of
[6] B. Hafizi and S. H. Gold, “Beam loading in magnicon deflection cavi-
neural networks for industrial optimization problems.
ties,” IEEE Trans. Plasma Sci., vol. 25, no. 1, pp. 53–59, Feb. 1997. Dr. Hemminger is a Vice-Chair of the Erie Section of the IEEE.
[7] S. Xiao and Y. Mo, “Study of open cavity filled with plasma density
grating,” IEEE Trans. Plasma Sci., vol. 27, no. 5, pp. 1495–1500, Oct.
1999.
[8] W. J. Vogel, H. Ling, and G. W. Torrence, “Fluorescent light interaction
with personal communication signals,” IEEE Trans. Commun., vol. 43,
no. 2/3/4, pp. 194–197, Feb./Mar./Apr. 1995.
[9] W. Elenbaas, Fluorescent Lamps, 2nd ed. London, U.K.: Macmillan,
1971.
[10] J. B. Murdoch, Illumination Engineering. New York: Macmillan, Matthew N. O. Sadiku (S’82–M’88–SM’90) was
1985. born at Shagamu, Nigeria on May 17, 1955. He
[11] J. J. Thomson, “On the discharge of electricity through exhausted tubes received the B.Sc. degree from Ahmadu Bello
without electrodes,” Phil. Mag., vol. 32, pp. 321–336, 1891. University, Zaria, Nigeria, in 1978, and the M.Sc.
[12] R. F. Harrington, Time Harmonic Electromagnetic Fields. New York: and Ph.D. degrees from Tennessee Technological
McGraw-Hill, 1961. University, Cookeville, TN, in 1982 and 1984,
respectively.
From 1984 to 1988, he was an Assistant Professor
at Florida Atlantic University, Boca Raton, where he
Sudarshan Rao Nelatury (M’00–SM’02) received did graduate work in computer science. From 1988
the B.Tech. degree from Jawaharlal Nehru Techno- to 2000, he was at Temple University, Philadelphia,
logical University, Hyderabad, India, in 1981 and the PA, where he became a Full Professor. From 2000 to 2002, he was with Lu-
M.E. and Ph.D. degrees from Osmania University, cent/Avaya, Holmdel, NJ as a System Engineer and with Boeing Satellite Sys-
Hyderabad, in 1985 and 1996, respectively. tems as a Senior Scientist. He is presently a Professor at Prairie View A&M
From 1983 to 1999, he taught at the Department University, Prairie View, TX. He is the author of over 130 professional pa-
of Electronics and Communication Engineering, pers and over 20 books including Elements of Eletromagnetics, 3rd ed. (Ox-
Osmania University. During 1999–2003, he was ford. U.K.: Oxford Univ. Press, 2001), Fundamentals of Electric Circuits with
a Visiting Faculty Member at the Electrical and C. Alexander, 2nd ed., (New York: McGraw-Hill, 2000), Numerical Techniques
Computer Engineering Department, Villanova in Electromagnetics 2nd ed., (Boca Raton: CRC Press, 2001), and Metropolitan
University, Villanova, PA. Presently, he is with the Area Networks (Boca Raton: CRC Press, 1995). Some of his books have been
Department of Electrical, Computer, and Software Engineering, at the School translated into Korean, Chinese (and Chinese Long Form in Taiwan), Italian,
of Engineering and Engineering Technology, Pennsylvania State University, Portuguese, and Spanish. His current research interests are in the areas of nu-
Behrend College, Erie. His research interests include electromagnetics and merical modeling of electromagnetic systems and computer communication net-
digital signal processing. works.
Dr. Nelatury is a Life Member of the Institution of Electronics and Telecom- Dr. Sadiku was the IEEE Region 2 Student Activities Committee Chairman.
munication Engineers (IETE) and the International Society for Technical Edu- He was an Associate Editor for IEEE TRANSACTIONS ON EDUCATION. He was
cation (ISTE), and a Member of American Society for Engineering Education the recipient of the 2000 McGraw-Hill/Jacob Millman Award for outstanding
(ASEE). contributions in the field of electrical engineering.