Download as pdf or txt
Download as pdf or txt
You are on page 1of 32

Home Search Collections Journals About Contact us My IOPscience

The anode region of electric arcs: a survey

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2010 J. Phys. D: Appl. Phys. 43 023001

(http://iopscience.iop.org/0022-3727/43/2/023001)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 141.225.218.75
The article was downloaded on 08/06/2012 at 21:32

Please note that terms and conditions apply.


IOP PUBLISHING JOURNAL OF PHYSICS D: APPLIED PHYSICS
J. Phys. D: Appl. Phys. 43 (2010) 023001 (31pp) doi:10.1088/0022-3727/43/2/023001

TOPICAL REVIEW

The anode region of electric arcs: a survey


J Heberlein1 , J Mentel2 and E Pfender1
1
Mechanical Engineering, University of Minnesota, 111 Church St. SE, Minneapolis, MN 55455, USA
2
Fakultät für Electrotechnik und Informationstechnik, Ruhr University, Universitätsstrasse 150, 44801
Bochum, Germany

Received 1 May 2009


Published 15 December 2009
Online at stacks.iop.org/JPhysD/43/023001

Abstract
Electric arc anodes frequently determine functional performance and lifetime of a number of
arcing devices, ranging from discharge lamps to plasma spray torches. While there have been
numerous studies of the anode region of electric arcs, our understanding of the detailed
physical processes is still limited. The reason for this lack of detailed knowledge is that
numerous factors influence the arc–anode interaction, and that the plasma–solid interface in
high intensity arcs is in general not accessible to diagnostics and one has to rely on indirect
measurements. Throughout this survey, the emphasis will be on high intensity arc anodes, i.e.
on plasmas with temperatures of more than 10 000 K and electron densities exceeding
1022 m−3 outside the boundary layer, and heat fluxes exceeding 107 W m−2 . The plasma
parameters in the boundary layer as obtained with different techniques by a number of
investigators for a variety of conditions are presented, and the effect of macroscopic flow
conditions is discussed. Experimental and modelling results are presented. A brief comparison
with low current arcs is also given, and the areas that need further research are highlighted.
(Some figures in this article are in colour only in the electronic version)

1. Introduction Research over the past 50 years provided a solid


understanding of the arc plasma including non-equilibrium
Electric arcs have been known for about 200 years and their effects, but the same is not true for the electrode regions
widespread and still growing application assured continuous which represent the interface and the electric contact between
research interest and demand over the years. There is a vast the relatively hot arc plasma and the electrodes. Results
range of currents used in arc applications, covering more than of earlier studies are summarized in a number of textbooks
five orders of magnitude. A typical example for an extremely and reviews [1–8] which indicate that the anode region did
high current application is the arc furnace used for steel not attract as much attention as the cathode region. On the
making with currents 100 kA. A wide variety of arc plasma other hand, anode erosion is frequently the lifetime controlling
torches have been developed for scrap melting, alloying, iron mechanism in non-transferred arc applications such as, for
melting in cupolas and remelting technologies with arc currents example, in plasma torches or arc lamps. In spite of research
typically in the 10 kA range. Plasma torches in the 100 A–1 kA efforts over the past 25 years [9–43], there are still gaps in our
range found extensive applications for plasma spraying and understanding of the anode region of arcs. One of the reasons
plasma synthesis. Classical arc welding employs free-burning for these gaps is the little understood interaction of the arc
arcs with currents in the range of 100 A to several hundred with the anode surface which is governed by a large number
amperes, and this current range is also found in arc cutting of effects, including those determined by the current and the
applications where the arc is extremely constricted. Finally, current density, and fluid dynamic effects. Different effects
arc lamps represent one of the most important arc applications dominate in different arc operating regimes, e.g. different arc
(about 25% of the electricity generated in the US is consumed currents and gas type and gas flow rates. Consequently, no
for lighting and arc lamps are the most efficient light sources) predictive model exists covering the entire range of possible arc
with currents typically in the 0.5–10 A range. parameters which are encountered in actual arc applications.

0022-3727/10/023001+31$30.00 1 © 2010 IOP Publishing Ltd Printed in the UKR


J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Low current arcs are mainly used in high intensity


discharge (HID) lamps, and the anode effects in such arcs are
discussed in an accompanying paper. The principal differences
between such arcs and high intensity arcs are that the anode
is usually not water-cooled, i.e. sufficiently hot for thermionic
electron emission. Furthermore, in the case of HID lamps, the
operating pressures are much higher, usually exceeding 1 MPa,
and there is a periodic change of the electrodes from anode to
cathode.
This survey will summarize our present knowledge of the
anode region for high intensity arcs (I > 50 A). The next
section will provide a qualitative description of the physical
effects in the arc anode region, followed by a description of
the influence of fluid dynamics. A review of experimental
studies is divided into sections describing different types of Figure 1. Different length scales in the anode region.
measurements. Results of model descriptions of the anode
region follow, and a section comparing the effects observed regions may be distinguished: (1) a region adjacent to the arc
with high intensity arcs to those obtained with low intensity column governed by increased conduction losses to the anode
arcs leads to the general conclusions. resulting in temperature non-equilibrium Te > Ta , (2) a region
of composition non-equilibrium due to diffusion fluxes towards
2. General characteristics of the anode region the anode surface and (3) the Langmuir space charge region
directly in front of the surface. Obviously, the first region
The anode boundary layer is the region in front of the anode overlaps with the other two regions and the second region
surface where the plasma properties, e.g. electron and heavy overlaps with the third region. The thickness of these regions
particle temperatures, electron densities, electric field strength, is strongly dependent on the fluid dynamic situation and they
are different from their values in the arc column. The anode may overlap; temperature and composition non-equilibrium
region is characterized by an extremely steep temperature may extend to 0.5–2 mm from the anode surface, while the
drop from plasma temperatures in the order of 104 K to anode space charge zone is typically confined to a region of less
temperatures in the order of or below 103 K, over a distance of than 0.05 mm.
1 mm or less, depending on pressure and flow conditions [14]. The physical processes in the anode region can be
A similar situation exists for the electron density which may explained with an energy balance on a control volume
drop from values in the order of 1023 m−3 in the plasma column element in the thermal and fluid boundary layer (where local
to values many orders of magnitude less at the anode surface thermodynamic equilibrium (LTE) is assumed to prevail), and
[9]. The interface between arc column and anode surface may with the generalized Ohm’s law for the non-equilibrium region,
be divided into the traditional boundary layer characterized by assuming rotational symmetry (with r the radial and z the axial
temperature and velocity gradients and strongly affected by coordinate):
the fluid flow situation, and into an electric boundary layer    
∂h ∂h 1 ∂ ∂T
consisting of a sheath and pre-sheath adjacent to the surface. ρ u +v − rκ
∂z ∂r r ∂r ∂r
The sheath is the location of net space charges, resulting in  
corresponding strong electric fields as described by the Poisson ∂ ∂T
+ κ + PR = j · E, (1)
equation, and it has an extent in the order of a few Debye ∂z ∂z
lengths [44].    
Figure 1 shows some characteristic lengths in the anode dne dTe dni dTi
j = σ E + kµe Te + ne − kµi Ti + ni
boundary layer, indicating that the sheath is essentially dz dz dz dz
collisionless for electrons; however, ion collisions, in (2)
particular, charge exchange collisions may occur in the
electrical boundary layer. The potential drop across the sheath with  R
is sometimes identified as the anode fall which is not quite I = 2π j rdr = const. (3)
correct. Although the potential drop across the sheath may o

be the main contribution to the anode fall, the thickness Here, ρ is the gas density, u and v are the axial and radial
of the anode fall region has to extend beyond the sheath velocity components of the gas, respectively, h is the gas
thickness to meet the various tasks of the anode fall [45–47]. enthalpy, T the gas temperature, κ the gas thermal conductivity,
Accordingly, the anode fall is defined here as the difference PR the volumetric radiation loss, j the current density and E
between the anode potential and the potential at the point where the electric field strength. In equation (2), σ is the electrical
the potential gradient changes from that in the arc column. This conductivity of the gas, k is the Boltzmann constant, µe and
change in electric field is the result of charge carrier losses and µi are the electron and ion mobilities, respectively, and Te and
energy losses to the anode surface. Following Nemchinsky and Ti and ne and ni are electron and ion temperatures and number
Perett’s description of the anode region [44], three different densities, respectively.

2
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 2. Illustration of anode voltage drops for diffuse and constricted attachments. uc is the potential at the column end and the dashed
line is an extrapolation of the column potential to the anode position, ua,d is the anode potential for a diffuse attachment with a negative
anode fall (ua,d < uc ), ua,c+ is the anode potential for a constricted attachment with a sheath potential drop and ua,c− is the anode potential
for a constricted attachment with a positive anode fall but a field reversal in the sheath region and a negative sheath potential drop. us
indicate the anode falls ua and the sheath potential drops us .

In the energy equation, the first term describes the on the area over which the plasma contacts the anode surface,
convective transport associated with macroscopic mass fluxes, and according to the density and temperature gradients which
the second and third terms describe thermal conduction in determine the current density (equation (2)), and it can be either
radial and axial direction, and the fourth term describes the negative, zero or slightly positive.
radiation loss. These energy losses are balanced by the A constricted attachment, in contrast will have first an
dissipation term on the right-hand side. In Ohm’s law, the first increase in the electric field due to the constriction; however,
term on the right-hand side describes the current flow due to very steep gradients immediately in front of the anode surface
charge carrier drift due to the electric field, while the remaining could lead to a field reversal. The composite of the voltage drop
terms describe the charge carrier flux due to diffusion due to in the constriction region and that in the region immediately
density and temperature gradients. in front of the anode will give the anode fall, which may be
If we consider an anode region without macroscopic flow, positive or negative. The various potential distributions in front
the first term in the energy equation can be neglected, and of the anode are illustrated in figure 2.
the increase in axial conduction losses must be met by an In addition to the influences of the fluid flow conditions,
increase in energy dissipation. Looking at Ohm’s law and chemical reactions on the anode surface as, for example,
the condition of constant current, this increase in dissipation encountered in arcs operated in atmospheric air or in other
requires a combination of increased electric field, decreased arc oxidizing fluids affect the anode attachment, and anode
radius and possibly increased diffusion fluxes. A decreased evaporation at a specific location is another mechanism which
radius, resulting in an increased current density, increases leads, in general, to spot formation [49, 50].
the self-magnetic pressure, resulting in a macroscopic mass Current continuity requires a sufficiently high electron flux
flow away from the anode [48], resulting in cold gas in- from the arc column to the anode surface leading to substantial
flow from the radial direction, enhancing the heat loss and deviation from LTE in the anode boundary layer [9, 44]. The
the necessary constriction (thermal pinch). Consequently, temperature of the heavy species approaches the temperature
macroscopic flows cannot be neglected in the treatment of of the anode in the immediate vicinity of the anode surface,
the anode boundary layer of high intensity arcs, and for the whereas the electron temperature remains sufficiently high
constricted attachment, the first term in equation (1) is an to ensure the required electrical conductivity as shown for a
additional energy loss term. typical example in figure 3 [9]. For the same conditions as in
If there is a superimposed flow towards the anode, e.g. figure 3, figure 4 [9] shows the electron density in the anode
because of a cathode jet, the plasma will form a stagnation boundary layer derived from a one-dimensional analysis.
layer over a larger region of the anode surface, i.e. an increase A comparison with electron densities under LTE
in the arc radius R and a radial outflow of plasma from the conditions reveals increasing discrepancies as the anode is
attachment zone. Consequently, the first term in equation (1) approached [9]. This is a clear indication that there are not only
represents a power increase in the energy balance, with a deviations from kinetic, but also deviations from composition
reduction of the energy dissipation term on the right-hand side equilibrium in the anode boundary layer. The actual electron
as a consequence. The electric field will adjust, depending density close to the anode may be more than a factor of two

3
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 5. Electric field and ion current density distribution in the


Figure 3. Calculated heavy particle and electron temperature anode boundary layer. (Reprinted from [9] copyright 1980
distribution in front of the anode surface (at left boundary). American Institute of Physics.)
(Reprinted from [9] copyright 1980 American Institute of Physics.)
3. Fluid dynamic effects

Fluid dynamic effects in high intensity arcs are either produced


by external flows superimposed to the arc or by flows induced
by the arc itself. The latter type of flow, induced by the
interaction of the arc current with the self-magnetic field, is
known as the Maecker effect [48]. Both types of flow may
exert a strong influence on the anode region. Throughout
this section, we will consider three different configurations:
a free-burning arc with a rod shaped cathode and an anode
perpendicular to the arc axis (see figure 6); a partially wall
stabilized arc, operating within a water-cooled cylindrical
channel, with a flat anode perpendicular to the arc axis and
with a gap between the channel assembly and the anode (see
figure 7); an arc between a rod cathode and the inside
of a cylindrical anode, serving as a nozzle for the flow
Figure 4. Electron density distributions in front of anode surface
calculated with Saha equation and considering diffusion fluxes. superimposed on the arc (see figure 8). These configurations
(Reprinted from [9] copyright 1980 American Institute of Physics.) cover the majority of the geometries used in praxis.

higher than the equilibrium density according to the predicted 3.1. Flow perpendicular to the anode
electron temperature. And even more important, the current Plasma flows directed towards the anode are encountered with
flow at the anode is primarily driven by electron diffusion. As the free-burning arc with a cathode jet and with the constricted
previously mentioned, this strong electron diffusion may even arc with a gap between the constrictor channel and the anode
lead to negative electric fields in front of the anode as shown and with superimposed flow towards the anode.
in figure 5 [9]. As discussed in the previous section, a flow directed
Figure 5 refers to the same conditions as figure 3. towards the anode will lead to electron density and temperature
The calculated electric field shows strong negative values gradients over a wide area, as well as to convective energy
immediately in front of the anode (sheath), and by calculating transport towards the anode. The steep gradients of density
the potential drop across this sheath one finds a value of −1.8 V and temperature in front of the anode, as shown by calculations
which may be identified as the direct anode or anode sheath and measurements in figure 9 [14], suggest that diffusion of
fall [9]. A negative anode fall implies that there is a positive electrons to the anode will be an important or even dominating
space charge in front of the anode and the negative electric transport mechanism [9]. The consequence is that an increase
field will drive an ion current to the anode, simultaneously in the electric field is avoided and a negative anode fall can be
decelerating electrons. This ion current, also shown in figure 5, observed.
has been confirmed by experiments which will be discussed in However, a reduction of the plasma flow towards the
section 4. In the case of a helium arc, a positive value for anode, e.g. by increasing the length of the arc and reducing
the direct anode fall has been reported resulting from similar any parallel flow, will reduce the convective transport towards
one-dimensional calculations [44]. the anode and lead to arc constriction. Under these conditions

4
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

3.2. Flow parallel to the anode


The situation of an arc exposed to an external flow parallel
to the anode is found in all types of arc plasma torches (see
figure 8). In addition to the externally superimposed flow, there
is also a self-induced flow in the same direction due to the
constriction of the arc at the cathode leading to the formation
of a cathode jet. Although the main arc is exposed to parallel
flow, the arc connection to the anode (anode column) is exposed
to cross flow. Wutzke et al [52, 53] analysed this situation
with an arc between a rod type cathode and two parallel plates
as anodes. If the superimposed flow rate surpassed a critical
value, the anode attachment became unstable and started to
move with a dark boundary layer between the arc and the anode
surfaces extending further and further downstream. Once the
voltage drop across this boundary layer at an upstream location
Figure 6. Schematic of free-burning arc. reaches a certain value, an upstream breakdown is observed
(restrike). The continuous repetition of this downstream
movement and restrike leads to the familiar sawtooth-like
voltage waveform. A linear relationship was found between
the arc current and the critical velocity at which the instability
started. While the forces acting on the anode attachment were
identified as the Lorentz force resulting from the changes in the
arc current density in the attachment region, and the drag force
of the cold gas acting on the low density attachment column
between the arc and the anode surface, a simple force balance
has not been able to describe the regions of arc stability [54].
The actual plasma torch with a cylindrical anode provides
a somewhat more complicated situation. From an analysis
of voltage traces obtained for such a torch as shown in
figure 8, Coudert et al [55] concluded that different types of
restrikes could occur, including downstream restrikes when
the boundary layer between the plasma jet and the anode
Figure 7. Schematic of wall stabilized arc with anode perpendicular downstream of a momentary attachment became very thin. A
to the arc axis and a gap between the constrictor and the anode.
correlation has been found between the attachment residence
time and the arc current and the mass flow rate using a non-
an increased field in the anode boundary layer and a positive dimensional analysis. Duan et al [56, 57] used the analysis
anode fall are expected even though a field reversal is predicted of the voltage traces to characterize the different modes of
due to the extremely steep gradients over the last few mean free the anode attachment: the ‘stable’ mode indicated by a steady
paths in front of the anode surface [32]. voltage, the ‘take-over’ mode characterized by random voltage
fluctuations and the ‘restrike’ mode with the sawtooth shaped
The dependence of the anode attachment on the plasma voltage trace. By using the values for the amplitude of
gas flow rate towards the anode for a geometry of a wall the voltage fluctuation, and for the slopes of the voltage
constricted arc with an anode perpendicular to the arc axis and traces during voltage increases and decreases, an algorithm
a 10 mm gap between the constrictor and the anode is shown was formulated describing the anode attachment stability for
in figure 10 [51]. any arc operating conditions, including ‘mixed’ modes, i.e.
It can be seen how increasing axial gas flow rates push modes showing characteristics of restrikes and of take-over.
the stagnation layer between the axial flow and the anode jet Assigning a mode value of 2 to a pure restrike mode, a value
towards the anode until the anode jet disappears and a diffuse of 1 to a pure takeover mode and a value of 0 to a steady
attachment exists. Figure 11 [51] shows the accompanying mode, the correlations were obtained shown in figure 13 [57],
changes in the arc voltage: a steep increase in the voltage indicating a monotonic decrease in mode value with increasing
current and decreasing mass flow rate.
before the transition to the diffuse mode followed by a
Through end-on observation of the arc inside the anode
voltage drop.
nozzle, a boundary layer thickness could be defined between
It should be noted that for arcs below 100 A, the lowest the luminous region of the arc and the anode surface. Figure 14
voltage is with a constricted attachment, while for arcs at higher [57] shows that there appears to be a clear correlation between
currents a diffuse attachment provides a lower voltage and a the boundary layer thickness and the mode value: a thicker
more stable operation. The stepwise change of the attachment boundary layer promotes a higher mode value (i.e. more
diameter with increasing axial gas flow is shown figure 12 [51]. restrike-like behaviour). Boundary layer thicknesses were

5
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 8. Schematic of a plasma torch with the flow parallel to the anode surface.

lateral gas flow parallel to the anode in the gap between the
constrictor and the anode (see figure 15). A gap between the
constrictor and the anode allows observation of the arc, and
cold gas can be introduced parallel to the anode surface. Inaba
et al [58] measured the deflection of the arc due to the lateral
gas flow and defined a ‘Young’s modulus’ and a stiffness for
the arc.
However, no physical explanation for the formulation was
offered. The same experiment was used to determine the effect
of the cross flow on the arc voltage [59] for an arc of up to 250 A
in argon with air as the lateral gas. The increase in voltage was
found to be uniquely dependent on the deflection of the arc root.
In a similar set-up and under similar conditions, Hartmann
and Heberlein [60] found that the arc attachment changes
from a diffuse attachment first to a constricted attachment,
possibly with several spots, then to a fluctuating constricted
attachment which is deflected from the original position, and
finally to a strongly deflected attachment with a restrike. The
thickness of the dark boundary layer outside the arc attachment
Figure 9. Measured and calculated temperature distributions in a region increases at first stepwise during the change from the
free-burning arc. (Reprinted from [14] copyright 1983 American diffuse mode to the constricted mode, then more slowly with
Institute of Physics.) increasing deflection until a maximum is reached and further
increases in lateral gas flow do not change the boundary layer
changed by changing the arc current and the gas flow rate. any more (see figure 16) [60]. Further experiments with
It is at present unclear why one set of data deviates from essentially the same experimental set-up with current values of
the correlation. 75 and 100 A and argon as the plasma gas and argon or nitrogen
As indicated in figure 8, there is a relatively thick cold as the lateral gas confirmed these observed effects of increasing
boundary layer between the main arc column and the anode lateral gas flow and found that nitrogen lateral gas flow will
surface and, as a consequence, the arc section across this result in stronger constriction [61, 62]. By using a segmented
boundary layer and the anode arc root will be constricted. With anode and correlating arc attachment position (obtained with a
increasing current, the main arc column expands, reducing the high speed video camera) with the current transfer and the heat
thickness of the anode boundary layer. As this boundary layer transfer to the different segments it was found that both current
shrinks, a more diffuse anode arc root becomes feasible. This and heat transfer followed the luminous attachment spot with
fact led to a puzzling observation of the anode life in arc plasma some current flow and heat transfer outside the spot [63].
torches. By increasing the arc current, the anode life of a Calorimetric and current measurements with a segmented
given torch may be enhanced, in spite of the fact that the heat anode show that never all the current is transferred to
transfer to the anode increases with increasing current (see downstream segments, but that a certain percentage of the
section 4.4). Anode melting and anode erosion, however, are current keeps flowing to the upstream anode segments with
determined by the specific heat fluxes rather than by the total increasing lateral gas flow, and that the heat transfer essentially
heat fluxes [12]. Specific heat fluxes are substantially higher follows the current flow (see figure 17) [63].
in the case of constricted anode arc roots (section 4.4). Further increase in the lateral gas flow rate will lead to arc
Several investigations of the anode boundary layer effects instabilities, first some random movement of the attachment
were made with the wall stabilized arc arrangement with a with voltage fluctuations similar to those observed with the

6
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 10. Photographs of arc attachment mode changes with different plasma gas flows; current: 100 A, argon 1 atm, constrictor–anode
gap:10 mm. (Reprinted from [51] copyright 2001 IOP Publishing.)

Figure 11. Voltage change during transition from constricted mode to diffuse mode. (Reprinted from [51] copyright 2001 IOP Publishing.)

‘take-over mode’ in plasma torches, and then to actual restrike • It appears that for low current arcs (<100 A) the
conditions, i.e. a sudden move to an upstream attachment constricted mode is more stable while for higher currents
spot. High speed movies showed that the frequency of the the diffuse mode provides lower voltages.
restrike movement jumped at a certain threshold lateral gas • Gas flow parallel to the anode surface favours a constricted
flow rate right away to 270 Hz, followed by a more gradual attachment and results in either a random oscillatory
increase. It was also observed that the location from which the movement of the attachment or in the periodic restrike
restrike occurred moved downstream with increasing lateral motion.
gas flow rate, i.e. the length of the restrike to the same
upstream position increases (see figure 18) [63]. Schlieren 4. Experimental characterization of the boundary
images of the arc attachment with lateral gas flow show that layer
the restrike is preceded by a fluid dynamic instability leading
to a reduction of the cold gas boundary layer thickness (see 4.1. Electron temperature and density measurements
figure 19) [64].
This section contains results of spectroscopic, Langmuir probe
While the initial increase in power dissipation during the
and laser scattering measurements in the anode boundary
constriction of the attachment is mostly transferred to the
layer.
anode, the increased power dissipation during the elongation of
the arc did not increase the heat transfer to the anode. One may (a) Spectroscopic measurements. There have been numerous
conclude that convective and radiative heat transfers are less spectroscopic measurements of the anode region of electric
significant than conduction and the heat transfer mechanisms arcs with the anode perpendicular to the arc axis. Free-burning
associated with current flow. arcs as well as wall stabilized arcs with a gap between the
The experimental observations may be summarized by the constrictor and the anode have been used. With free-burning
following statements: arcs, the principal experimental parameters that influence
the anode attachment are the arc current and the anode–
• The fluid flow influences the thickness of the thermal cathode gap. With the constricted arc, the anode attachment
boundary layer in front of the anode and consequently is influenced by the arc current and the gas flow parallel to
the physical processes in the arc anode attachment. the arc towards the anode. The latter arrangement has also
• Two distinctly different attachment modes have been been used to simulate anodes with the arc parallel to the anode
observed, depending on the boundary layer thickness, the surface by introducing a cold gas flow parallel to the anode
constricted and the diffuse attachment. surface towards the arc attachment. In all cases reported here,

7
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

anode erosion and a strongly constricted anode attachment Heberlein and Pfender [65] report spectroscopic measure-
with anode melting have been avoided. Furthermore, almost ments with a 100 A argon arc using a constrictor assembly with
all investigations have been carried out in argon at pressures a gap of 3.5 mm between the constrictor and the anode. Results
between 80 and 110 kPa, and deviations from these conditions were obtained for a diffuse attachment with the assumption of
are noted. The difficulty with spectroscopic measurements is PLTE, and measurements of several absolute line intensities
the existence of non-equilibrium in the boundary layer, and and continuum intensities yielded electron temperature and
different approaches have been chosen to obtain results for density values at different axial locations (see table 1). Elec-
non-equilibrium conditions. tron density values of 5 × 1022 m−3 or 6 × 1022 m−3 were ob-
Dyuzhev et al [30] report results of spectroscopic tained using the PLTE assumption or the continuum coefficient,
investigations of a free-burning arc in which the anode was a respectively, at 2 mm from the anode, and of 0.9 × 1022 m−3
rotating water-cooled copper cylinder. The current was varied and 1.2 × 1022 m−3 at 0.5 mm from the anode. For a similar
over a wide range, and reducing the current below 50 A led to set-up but with a gap of 10 mm and an arc current of 200 A,
a change in the anode attachment from a diffuse attachment Sanders et al [12] report similar values (see table 1). A com-
to one consisting of several parallel branches, and eventually, parison of temperature profiles for a diffuse and a constricted
at a current of 15 A, to a single constricted attachment. The attachment showed similar values at a location of 1 mm in front
electron density data derived from continuum measurements of the anode; however, with the constricted attachment a tem-
and assuming partial local thermodynamic equilibrium (PLTE) perature maximum of 12 400 K approximately 2 mm from the
are 1.9×1022 m−3 at 0.5 mm in front of the anode for the diffuse anode was found. These results lead to the conclusion that
attachment, and for the constricted attachment, an electron beyond 2 mm from the anode LTE can be assumed without
density maximum is observed close to 1023 m−3 at about 1 mm causing too much of an error, and that the temperatures are in
in front of the anode. the order of 12 000 K, a value which is supported by modelling
The most complete measurements for free-burning arcs results.
have been reported by Tanaka and Ushio [36–38] for an arc Recent Thomson scattering measurements with a wall
gap of 5 mm and arc currents ranging from 50 to 150 A. The stabilized arc and a gap in front of the anode resulted in
assumption of LTE existence has been avoided by using laser electron temperature and density distributions in the entire
scattering measurements, i.e. Rayleigh scattering for atom anode region between 0.05 and 5 mm from the surface at 1 atm
temperatures and Thomson scattering for electron temperature in argon for currents between 60 and 120 A [64, 66]. Variation
measurements. Rayleigh scattering was performed up to of current and parallel gas flow allowed a controlled change
distances of 50 µm from the anode with a roof type anode of the anode attachment type, and two transitional types of
arrangement sloping towards the observer. Langmuir probe attachment between the diffuse attachment and the constricted
measurements with a probe embedded in the anode provided attachment have been observed when the parallel gas flow was
additional data on electron temperatures. The results are given reduced. Figures 20 and 21 show the electron temperature and
in tables 1 and 2. If we consider the action of the cathode electron density distributions along the arc axis for varying
jet and the proximity of the cathode to the anode (5 mm), the currents and axial mass flow rates [66].
results appear to be consistent with the other measurements, For diffuse attachments (high axial gas flow rates) electron
except for the very high electron temperature value measured temperatures and densities increase strongly with current at
with a Langmuir probe at the anode surface for the 50 A arc. both locations, 2 and 0.05 mm from the anode surface. When
The effect of the anode state has been investigated by the axial gas flow rate is decreased, electron temperatures and
Etemadi and Pfender [15] who compared temperature profiles densities initially decrease. However, at a certain flow rate a
of a 150 A free-burning argon arc with a 10 mm gap, 1 mm in transition is observed to a constricted attachment with higher
front of a solid copper and a molten copper anode. A drop electron temperatures and densities, increasing at 0.05 mm in
in peak LTE temperature from 13 000 to about 12 100 K was front of the surface from 10 900 K with 5 slm to 14 400 K
found when copper vapour was present, even though no strong at 0.7 slm, and from 0.8 × 1022 m−3 to 3.8 × 1022 m−3 ,
constriction was observed. Gleizes et al [33] extended these respectively. Both distributions show maxima in a region
measurements to different anode materials, copper, iron and between 2 and 0.05 mm from the anode. The change from
steel, all in their solid and molten forms, with a 90 A argon arc. diffuse to constricted attachment is preceded by the formation
At a location 1 mm in front of the anode, LTE temperatures of a cold boundary layer between the arc plasma and the anode
of 11 500 K were measured for Cu anodes, compared with surface in the arc fringes (lift-up mode) and by an increase in
10 750 K for steel anodes. For liquid anodes with carefully current density in the central region of the attachment. This
controlled evaporation to avoid arc constriction, the data were mode is followed by the formation of several attachment spots
9600 K for a steel anode and 9000 K for the molten copper at different locations on the arc periphery (multiple attachment
anode. The lower temperatures in all cases of metal vapour mode). A more detailed description of this transition will be
presence are explained by higher radiation losses and wider presented in section 4.3.
arcs because of the higher electrical conductivity of the metal The same arrangement has been used to characterize the
containing plasma at lower temperatures in the arc fringes. attachment with cold gas flow parallel to the anode surface,
The differences of the results with different anode materials simulating the parallel flow situation [63, 64]. Langmuir probe
were explained by different arc root diameters due to different and Thomson scattering measurements have been performed
thermal conductivities of the anode materials. for varying lateral gas flows, and the symmetric attachment

8
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 12. Stepwise change of anode attachment diameter with plasma gas flow during the transition from constricted to diffuse attachment.
(Reprinted from [51] copyright 2001 IOP Publishing.)

Figure 13. Arcing mode for arc with flow parallel to the anode, for
different currents and different gas flow rates. (Reprinted from [57]
copyright 2002 ASM International.)
Figure 15. Schematic of wall stabilized arc with perpendicular
anode and lateral gas flow. (Reprinted from [63] copyright 2006
IOP Publishing.)

has been observed to become asymmetric with a higher


temperature gradient towards the cold flow, followed by
a movement of the attachment from the arc axis in the
downstream direction, with an eventual upstream restrike.
Thomson scattering measurements show that the restrike is
preceded by a high electron temperature cloud at the location
where the restrike will occur [64].
One may conclude from these reported results that for a
wall stabilized arc, the non-equilibrium boundary layer region
extends from the anode surface to a location of 1–2 mm in
front of it, depending on the flow configuration. Most results
indicate that the heavy particle temperatures drop rapidly
towards the anode surface, while the electron temperatures
remain high showing only a slight drop. In short free-burning
Figure 14. Dependence of arc mode on the boundary layer arcs with arcing gaps of 5 mm or less, it is unclear where
thickness. cathode jet non-equilibrium ends and anode boundary layer

9
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 16. Thickness of anode boundary layer as function of lateral (sheath) gas flow [60]; current 100 A, argon 1 atm, plasma gas flow
5 slm, constrictor–anode gap 10 mm.

Figure 17. Fraction of heat (top) and current (bottom) transferred to


different anode segments as function of lateral gas flow rate. Figure 18. Restrike frequency (top) and distance (bottom) for
(Reprinted from [63] copyright 2006 IOP Publishing.) different lateral gas flow rates. (Reprinted from [63] copyright 2006
IOP Publishing.)

non-equilibrium starts [67]. Higher temperature gradients are


a tungsten wire surrounded by an insulating sleeve, embedded
seen in the anode boundary layer of these arcs. In particular the
in the anode. In most experiments, the probe was flush with
extremely high heavy particle temperature gradients derived
the anode surface; however in a few experiments, the probe
from Tanaka and Ushio’s measurements (about 9000 K over
tip was raised up to 0.25 mm above the anode surface [68].
50 µm) pose questions if heavy particle temperature continuity
In all experiments, the arc was operated in a stainless steel,
can be assumed, or if a temperature slip condition exists at the
vacuum tight chamber to ensure a pure argon atmosphere. The
anode surface. The anode material does affect the size of the
pressure in the chamber is maintained at 100 kPa. Special
attachment and consequently the electron and heavy particle
precautions were taken to prevent contamination of the probe
temperature in the boundary layer; however, no data are
surface during measurements [68].
available on the effect of the non-equilibrium due to different
anode materials. According to the spectroscopic measurements, we expect
for the measured electron temperatures and densities a Debye
(b) Langmuir probe measurements. Langmuir probe length of about 0.1–1 µm (depending on the exact value of the
measurements have been performed for both arrangements electron density), an electron mean free path of 10–20 µm and
discussed in the previous sections, with a wall stabilized arc and an ion mean free path of about 0.5 µm. For determining the
a gap between the constrictor channel end and the anode, and electron temperatures from the electron retarding regime of
with free-burning arcs. Measurements have been obtained for the probe characteristic, one can therefore use the collisionless
diffuse and for constricted attachments. The probe consisted of probe theory. However, it is more difficult to obtain values for

10
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 19. Development of fluid dynamic instability leading to restrike [64]. The dark band indicates a high electron density gradient. The
white line is drawn into the figure at the border of the dark band to illustrate the arc boundary, and the anode surface is indicated as a straight
line. (Reprinted from [64] copyright 2007 IOP Publishing.)

Table 1. Results of spectroscopic measurements (all measurements in Ar at or close to 1 atm).


Current Distance Temperature
Configuration (A) from anode (mm) Method (K) Reference
Wall stab. 100 2 Spectr., 11 500 [65]
0.5 PLTE 9 200
Wall stab. 200 1 Spectr., LTE 12 200–12 400a [12]
Wall stab. 60–120 0.05 Thomson 9 000–12 400 [66]
2 scatt. (diff)
11 000–13 450
Wall stab. 100 0.05 Thomson 14 400 (constr.) [66]
2 scat. 14 350
Free burning 150 1 Spectr, LTE 12 900b –12 400 [15]
(10 mm)
Free burning 90 1 Spectr., LTE 11 500 Cuc [33]
(18 mm) 10 750 steel
Free 50 1 Thomson 12 200 (Te ) [37]
burning 150 1 scatt. 17 000 (Te )
(5 mm) 50 1 Thomson 8 000 (Tn )
150 1 scatt. 13 000 (Tn )
Rayleigh scatt.
Rayleigh scatt.
Free 50 0.05 Rayleigh 1 500 [36]
burning 75 0.05 scatt. 2 500
(5 mm) 150 0.05 10 000
a
A slightly higher temperature value was observed for a constricted attachment.
b
Two different Cu anodes were used, one regular and one molten and resolidified.
c
Different anode materials were used, Cu giving the highest temperature value.

the electron density because in the electron saturation regime, anode [68]. The straight line in this semi-logarithmic plot
the probe will become the anode, and in the ion saturation is an indication that the electrons retain a MB distribution
regime one has to consider collisions in the sheath. throughout the boundary layer.
Operating the probe in the electron retarding regime and Figure 23 shows electron temperatures at the anode
assuming that the electrons have a Maxwell–Boltzmann (MB) surface as a function of the arc current with the electrode gap
distribution, the current density, je , collected by the probe will as parameter [68]. The electron temperature throughout the
allow derivation of electron temperature values. boundary layer up to the anode surface stays above 104 K. It
Figure 22 shows an example of a Langmuir probe should be pointed out that both an increase in the arc current
characteristic with the probe biased with respect to the and a decrease in the electrode gap enhance the velocity of the

11
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Table 2. Results of Langmuir probe measurements in anode boundary layer (all measurements in Ar at or close to 1 atm).
Arc Te Te Te Te Te Te Te
Configuration current (A) z = 0 µm 50 µm 100 µm 125 µm 150 µm 200 µm 250 µm Reference
Wall 100 9 200 9 700 10 100 [13]
stab.— 150 8 800 9 700
diffuse 200 8 700 9 600 10 100
150 9 100 10 400 12 100
constricted 200 9 840 10 600 10 100
12 100
Wall stab. 80 9 600 [63]
100 9 200
80 10 250
Free 50 23 000 15 000 13 000 [36]
burning 100 16 000
(5 mm) 150 11 000 14 000 14 000
11 300 12 000
Free 75 11 200–11 400 [68]
burning 100
(7 mm) 150 12 900–13 200
14 000
Free 75 11 300 [68]
burning 100 12 600
(10 mm) 150 14 000

Figure 20. Axial electron temperature (a) and electron density (b) Figure 21. Axial electron temperature (a) and density (b)
distributions in the anode boundary layer for four different arc distributions in front of the anode for different axial gas flow rates.
currents; argon, 1 bar. The anode surface is on the left hand side. (Reprinted from [66] copyright 2007 IOP Publishing.)
(Reprinted from [66] copyright 2007 IOP Publishing.)

12
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

a strong increase in the electron temperature values at the


anode surface when the current is decreased from 150 to
50 A (23 000 K at 50 A). While they explain the temperature
increase with an increasing negative space charge and a positive
anode fall for a 50 A arc current, it cannot be explained why
measurements under comparable conditions [13, 68] do not
show these high temperatures. It should be mentioned that
the anode in the experiments by Tanaka and Ushio was not
planar but raised towards the centre, which could change the
attachment conditions.
Very recent Langmuir probe measurements in which the
boundary layer thickness could be controlled by lateral gas flow
parallel to the anode surface in the gap between a constrictor
and the anode [62, 63] show similar values as those reported
Figure 22. Langmuir probe data for deriving electron by Sanders and Pfender [13] and by Leveroni-Calvi [68]. It is
temperatures [68]. interesting to note that electron temperature values with lateral
gas flow, at a location downstream of the arc attachment, are
lower than in the attachment region, while they are higher
upstream of the attachment, in particular when the attachment
is deflected from the axis due to the lateral gas flow. A
strong initial constriction is observed when nitrogen is used as
lateral gas accompanied by an electron temperature increase
of 2000 K to about 11 800 K.
In summary, all measurements show high electron
temperatures up to the anode surface with values at about
or above 9000 K. However, there are strong differences in
the electron temperature characteristics for free-burning arcs
and wall constricted arcs. Free-burning arcs with a short gap
(5–7 mm), i.e. a cathode jet domination, show a strong increase
in the electron temperatures with current, i.e. with the jet
action according to Leveroni-Calvi [68], or a strong decrease
with increasing current according to Tanaka and Ushio [36].
Figure 23. Electron temperature values at the anode derived from For wall stabilized arcs with moderate gas flows, a weak
Langmuir probe data as function of current and arcing gap [68]. constriction can be assumed, and the data by Sanders and
Pfender [13] and by Yang and Heberlein [66] are in agreement,
cathode jet in front of the anode, i.e. they reduce the thickness showing a decrease in the electron temperature with increasing
of the anode boundary layer, resulting in a broader stagnation arc current. With this type of attachment, an increase in arc
region and a larger contact area between the plasma and the current results in an increase in the attachment diameter. This
anode. behaviour is in contrast to a strongly constricted attachment
Probe measurements on free-burning arcs and on wall where a strong increase in electron temperature is noticed [66].
constricted arcs have been performed by Sanders and Pfender Most measurements show an increase in temperature with
[13] (see table 2), and electron temperatures have been derived increasing distance from the anode surface, and reasonable
for diffuse and for constricted anode attachments and for agreement with values derived from emission spectroscopy.
currents ranging from 100 to 250 A. In the case of free-burning However, the probe measurements by Tanaka and Ushio show
arcs, the attachment was changed by increasing the arc gap a strong decrease in the electron temperature from the anode
(larger gaps leading to transitions to the constricted mode), towards the arc column for diffuse attachment conditions. This
and with the wall constricted arcs by changing the axial gas discrepancy in the reported data cannot be explained at present.
flow rate. Measurements were taken flush with the anode,
and with probe positions 0.125 and 0.25 mm above the surface 4.2. Current density measurements
(the side walls of the probe were insulated). For the diffuse
mode, the electron temperature at the surface decreased with Langmuir probes can also be used to measure current densities
increasing current, but remained constant at a higher value by measuring the current when the probe is at anode potential.
above the surface. For the constricted attachment, electron Because of the small probe diameter, it is permissible to divide
temperature values initially increased with increasing current, measured probe currents by the probe area to obtain current
but then appeared to remain steady or decline. The values also densities. Typical results are shown in figure 24 [68] for a
increased with distance from the anode, as seen in table 2. free-burning arc with arcing gaps between 7 and 15 mm. The
The Langmuir probe measurements performed by Tanaka results indicate a clear non-monotonic dependence on the arc
and Ushio [36] have already been mentioned. They report current, displaying a minimum which, for shorter arc gaps, is

13
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 26. Fraction of electron current arriving at anode surface,


Figure 24. Electron flux data derived from Langmuir probe data for derived from Langmuir probe measurements for different arc
different arc currents and arc gaps [68]. currents and arc gaps; lower currents and larger gaps show a more
constricted attachment [68].

arcs, the current densities derived from spectroscopic


measurements for the wall constricted arcs in front of the
anode show significantly lower values than those shown in
figure 24. Heberlein and Pfender [65] report 200 A cm−2 ,
and Sanders and Pfender [13] measured 100 A cm−2 for
arc currents of 100 A, compared with 500–1400 A cm−2
(depending on arc gap) in the free-burning arc measurements.
The current densities derived from Langmuir probe and
Thomson scattering measurements by Yang et al [63, 66]
range from 120 to 160 A cm−2 with increasing degrees of
constriction, and exceed 500 A cm−2 for a fully constricted
attachment. In general, higher electron temperatures correlate
with higher current densities.
Figure 25. Ion flux data derived from Langmuir probe
measurements for different arc currents and arc gaps [68]. 4.3. Anode fall measurements

very pronounced. The increase in je for arc currents, below Determination of the magnitude of the anode fall has been
the value at which the minimum occurs, is due to a more of strong interest to researchers for a long time, because
constricted arc root which is associated with a weaker cathode knowledge of the electric field distribution in the boundary
jet at these lower currents. The arc constriction at the anode layer in front of the anode will not only enhance our
at low arc currents appears to be more severe with narrower understanding of the physical processes in this region, but it
arc gaps. As figure 24 indicates, the current density minimum will also facilitate control of the anode attachment which is
shifts to higher arc currents for larger electrode gaps which is very important for the control of many plasma processes.
due to the weakened effect of the cathode jet at larger gaps [68]. Busz-Peuckert and Finkelnburg [69, 70] measured the
As previously mentioned, for negative anode falls there potential of a tungsten probe swept across the arc 1 mm in front
must be a positive space charge in front of the anode. The of the anode using a free-burning arc, for arc gaps ranging from
reversed electric field as well as strong ion density gradients 2 to 10 mm and arc currents from 10 to 200 A. For all distances,
in front of the anode will drive a positive ion current to constricted attachments were observed at arc currents below
the anode as shown in figure 25 [68]. Although this ion 25 A and diffuse attachments for currents above 75 A. They
current is below or close to 1% of the total arc current for equated the potential difference between the floating potential
arc currents below 100 A, the percentage of ion current to the of the probe and the anode potential with the anode fall. The
total current increases substantially with increasing arc current results are shown in table 3. The data for the fall increased
and decreasing electrode gap. Figure 26 [68] shows the ratio of with increasing arcing gap (lower cathode jet velocity) and
electron to ion current densities as a function of arc current with decreased with increasing arc current (increased cathode jet
the electrode gap as parameter. There is a dramatic reduction velocity). The anode fall values reflect the potential drop across
of this ratio with increasing arc current or equivalent, with the boundary layer and the sheath region, but no correction was
increasing attachment diameter. used for the probe sheath. Of interest is the comment that a
Similar to the differences in electron temperature data fall of 0 V should be expected for high ion flow into the anode
between results from free-burning arcs and wall constricted boundary layer.

14
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Table 3. Results of sheath potential measurements (all measurements in Ar at or close to 1 atm).


Distance from
Configuration Method anode (mm) I (A) U (V) Reference
Free-burning Probe swept 1 25 12 [70]
arc (10 mm) through arc
Free-burning Probe swept 1 25 7.5 [70]
arc (2 mm) through arc 1 200 4
Free-burning Energy 50 5.9 [71]
arc (6 mm) balance 100 4.5
Free-burning Energy 100 4.5 [72]
arc (6 mm) balance 150 3
Free-burning Gap change from 10 3 [74]
arc 4 mm–0 mm
Free-burning Probe 0, 0.25 100 −2.44, −3.4 [13]
arc/wall stab. in anode 150 −2.1, −3.1
arc— 200 −2.1, −2.7
Constricted
0, 0.25 150 −2, −3.87
Diffuse 200 −2.35, −4.0
Free-burning Probe 0 60 −3 [68]
arc (10 mm) in anode 100 −3 to −4
Wall Thomson 0.05 to 2 100 −0.7 diffuse [66]
stabilized arc scattering −0.1 in lift off

Schoeck and Eckert [71] and Schoeck and Maisenhälder high electron densities in front of the anode will result in high
[72] derived anode fall values from an energy balance of the diffusion currents and eliminate the need for ion production in
anode of a free-burning arc for pressures from 1 to 50 bar. the anode sheath. It must be noted that these MPD discharges
The heat transfer to the anode was assumed to consist of are quite different from arc discharges with quite high ion
convection and radiation from the arc column, and of the terms temperatures, and that the pressure at the anode is not given.
associated with the current transport (see section 4.4), namely But the fall value for a current density of 100 A cm−2 is
electron condensation, electron enthalpy flow and electron comparable to that reported by Busz-Peuckert and Finkelnburg
energy gained in the fall region. Values of 4.5 V at 100 A [70] for similar current densities.
and 3 V at 150 A are reported, and the values decrease with Several of the cited publications mention that anode
increasing pressure. Convective and radiative heat transfer fall values of zero or even negative fall values should be
contributions were determined through calculations using possible; however, all report positive voltage drops in the
simplifying assumptions, and relatively minor errors in these anode boundary layer. In contrast, Sanders and Pfender [13]
values could significantly change the anode fall values. and Leveroni-Calvi [68] report negative anode fall values for
A recent study by Hemmi et al [73] also derives atmospheric pressure argon arcs under all conditions they have
electrode fall values from electrode energy balances, i.e. considered. Sanders and Pfender used the Langmuir probe
from material evaporation rates from un-cooled electrodes. arrangements described in section 4.1, the probe tip being
Heat conduction/convection and radiation from the arc are either flush with the anode surface or protruding 0.25 mm
neglected. For a silver anode and an air arc with one half wave into the plasma. Both free-burning arcs and wall stabilized
of an ac current, with peak values varied from 350 to 1200 A, arcs were investigated, and the conditions were adjusted to get
an anode fall voltage of about 4.5 V is found, independent of either a diffuse or a constricted attachment. The floating probe
peak arc current. potential Vf was corrected to give the plasma potential Vs in
Dickson and von Engel [74] measured the voltage change front of the anode using a collisionless sheath assumption and
of a free-burning arc when the anode approached the cathode. neglecting the ion temperature difference:
Two steps are visible in the voltage trace, the first one was
associated with the collapse of the anode sheath and merging Vs = Vf + (kTe /2e) ln(2mi /π me ). (4)
with the cathode sheath (the anode fall voltage), the second
one when the two electrodes met (the cathode fall voltage). Electron temperature values derived from Langmuir probe
Different values are reported for different anode materials, with measurements were used. Two voltage drops were
a value of 3 V for a 10 A argon arc between copper electrodes. distinguished, the drop across the boundary layer derived from
Values for a nitrogen arc were significantly higher (∼10 V). the probe measurements at 0.25 mm in front of the anode and
Oberth and Jahn [75] determined fall voltages in high the ‘direct anode fall’ measured with the probe flush with the
current MPD arcs with a potential probe 1 mm in front of the anode surface, i.e. the drop across the collisionless sheath. The
anode surface. They corrected the measured potential for the results are shown in table 3. We see, that
sheath drop, and determined the current density distribution (1) in all cases the direct fall voltage as well as the total fall
with magnetic probes. Anode fall values of 10–12 V for voltage are negative;
current densities of 100 A cm−2 and of approximately zero at (2) the direct fall voltages do not vary strongly between
900 A cm−2 are reported. The point was made that sufficiently a diffuse and a constricted mode; however, the total

15
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

fall voltages are higher (more negative) for a diffuse a diffuse and a lift-up mode attachment for a 100 A argon
attachment, indicating a stronger positive space charge arc. The distributions of the electric potential parallel to the
in the boundary layer in front of the sheath; arc axis at three different radial locations are also shown, at
(3) the fall voltage increases with increasing current the centreline, r = 3 mm and r = 5 mm (6 mm for lift-up
(becoming more negative) for the diffuse attachment while mode). It can be seen that for the diffuse mode, a field reversal
it decreases with increasing current for the constricted (potential maximum) occurs at about 2 mm from the anode
attachment; surface, and slightly closer at larger radii. It also must be
(4) in the diffuse attachment, about half of the (negative) noticed that strong radial field gradients exist. For the lift-
anode voltage drop occurs in the sheath, while the other up mode, a second field reversal is visible at the r = 6 mm
half occurs in the boundary layer in front of it, whereas location approximately 0.4 mm in front of the anode, resulting
in the constricted attachment, the majority of the drop in a positive sheath potential drop. Electric field distributions
occurs in the sheath. The potential drop in the boundary in constricted attachments show strong radial variations of the
layer raises the question whether in these experiments a field, but slightly positive slopes as well as slightly negative
constriction actually existed. slopes are seen in the attachment region. Figure 28 [66]
summarizes the axial potential distributions for the different
Leveroni and Pfender [76] and Leveroni-Calvi [68] attachment modes.
present results of measurements with a Langmuir probe Based on these data, one may arrive at the following
embedded in the anode of a free-burning arc, and the boundary conclusions:
layer thickness was varied by changing the arcing gap between
(1) In a diffuse attachment a negative potential drop towards
7 and 15 mm and the current from 60 to 250 A. The voltage drop
the anode surface exists over the entire attachment region;
across the boundary layer, i.e. across the region of a significant
there is sufficient energy transport from the arc column
temperature drop, is derived by considering diffusion fluxes in
into the anode region to maintain a large enough contact
the analysis. The values are shown in table 3. These values
area of the plasma with the anode surface to carry the
are comparable to those reported by Sanders and Pfender [13].
current.
However, in both cases a collisionless sheath was assumed for (2) Reducing the energy flux from the arc column, the heat
the derivation of plasma potential values from the measured loss to the anode first manifests itself in a reduced high
probe potential data. While the Debye length for the described temperature region in front of the anode; while the current
conditions is shorter than the electron mean free path, the mean density in the attachment region increases, it is insufficient
free path for the ion–neutral charge exchange is shorter than to carry the current, and in order to transport the current
the Debye length. At present it is unclear how collision effects through the cooler fringe regions, a positive potential drop
will change the presented data. is needed.
Comparison of the temperature measurements and the (3) An instability analysis indicates [77] that a positive
potential measurements lead to a few definite conclusions: potential drop can lead to an electron-thermal instability,
(1) The fluid flow in the anode boundary layer affects the which will result in multiple constricted attachments and
temperatures as well as the potential distributions; probes local anode heating and evaporation in the arc fringes
mounted flush with the anode show a slight decrease in where the positive potential drop is initially seen; these
electron temperature and in the potential drop across the local spots surrounding the attachment have been observed
sheath with increasing current for a diffuse attachment, in experiments [78].
and an increase with increasing current for the constricted (4) A constricted attachment is characterized by an increase
attachment, very likely primarily following changes in the in electron temperature in the region immediately in front
current density. of the anode, and by a maximum in the electron density
(2) While there is for the most part an agreement between the distribution close to the anode, resulting in steeper electron
electron temperature measurements by different authors, density gradients and likely also in steeper ion density
significant differences exist between the different potential gradients. While these results should indicate an even
measurements. Agreement exists only in so far that higher fraction of arc current being driven by electron
the potential drop in the sheath is negative for a diffuse diffusion, the ion flux would reduce the total current. A
attachment as measured by probes flush with the anode positive anode fall voltage can exist increasing the ion
surface. current contribution to the total current by reducing the ion
flux towards the anode, and by accelerating the electrons
Indirect determination of the potential distribution in the anode towards the anode. However, the steep radial gradients
boundary layer has been pursued by Yang and Heberlein [66]. of electron density and temperature result in electric field
The Thomson scattering data have been used to determine distributions strongly distorted by diffusion fluxes, and
the distributions of electron partial pressure and electron a one-dimensional analysis of the constriction region is
temperature in the boundary layer. Neglecting the ion current likely not appropriate.
and the thermodiffusion current, an estimate of the electric field (5) In the case of a highly constricted attachment, the electron
has been obtained using equation (2) and current conservation saturation current density may be approached (in the order
for the different attachment modes. of 108 A m−2 ), and the ion flux away from the anode is
Figure 27 [66] shows maps of the electric field magnitude required to transport the total current. A positive sheath
and direction on the right-hand side of the anode region for potential should be the consequence.

16
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 27. Electric field distributions in the right half of the anode boundary layer of an 100 A argon arc (top); the arc axis is on the
left-hand side and the anode surface is at z = 0. The potential values as function of distance from the anode surface are shown below for
three different radial locations. (Reprinted from [66] copyright 2007 IOP Publishing.)

(6) Very steep ion temperature gradients in a strongly


constricted attachment will decrease the ion current
contribution to the total current because of a stronger
increase in ion diffusion fluxes towards the anode due to
the temperature gradients (see equation (2)). In contrast,
the electron temperature gradients remain low or are even
negative. This increased ion flux towards the anode may
require a positive potential drop to reduce the ion flux and
assure current continuity. An atomistic model is required
to determine the dominant effects under these conditions.

4.4. Heat flux measurements


Anode heat transfer and the distribution of heat fluxes at the
anode of high intensity arcs is, to a large degree, determined
by the properties of the anode boundary layer, in particular,
by its thickness. High anode specific heat fluxes may be
Figure 28. Axial potential distribution for a 100 A argon arc at
detrimental for the lifetime of anodes in dc arc devices, different attachment modes indicating positive anode falls for the
but they are very important for arc welding and plasma constricted and multiple attachment modes, and a negative anode
cutting and for some applications in plasma metallurgy [79]. fall for the diffuse mode. (Reprinted from [66] copyright 2007 IOP
Because of the significance of anode heat transfer for arc Publishing.)
applications, there have been numerous anode heat flux
measurements performed. Most of these measurements have transfer (assuming LTE). The difference between the measured
been calorimetric total heat transfer measurements, coupled
heat transfer and the calculated values were attributed to the
with some additional characterizations yielding information
energy transfer associated with the anode fall. At 100 A and a
on the different contributions to the energy flux to the anode.
pressure of approximately 200 kPa, the total heat transfer rate
Schoeck and Eckert [71] and Schoeck and Maisenhälder
[72] determined calorimetrically the total heat flux and spectro- was about 1.4 kW, with about 38% attributed to electron en-
scopically the temperature distribution for a free-burning argon thalpy flow, and 31% to the anode fall heating, about 21% to
arc. The temperature distribution allowed calculation of the convection and 10% to radiation.
radiative heat transfer and, with an assumed heat transfer coef- Sanders and Pfender [13] used a somewhat similar
ficient, the convective heat transfer, and the electron enthalpy approach; however, they separated heat conduction and heat

17
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

convection terms. The convection term was determined by


measuring the difference in total heat transfer for the cases
of a constricted attachment (with anode jet) and of a diffuse
attachment. The contributions associated with the electron
flux were calculated from electron temperature measurements
using a Langmuir probe, the radiative flux was estimated from
published values and the conduction flux was determined as
the difference between the individually determined fluxes and
the total measured flux value. Total fluxes were measured
of approximately 1 kW and 2 kW for an atmospheric pressure
argon arc at 100 A and 200 A, respectively. The individual
contributions in the constricted attachment case were 83% from
electron flux associated energy transfer mechanisms, 5% from
radiation, 12% from conduction. For the diffuse attachment the
contributions were 51% from the electron flux, 5% radiation,
11% conduction and 33% convection (i.e. all of the additional
heat transfer). It must be pointed out that the total current was
the same for the constricted and the diffuse attachment. The
strong increase in total anode heat flux in the diffuse mode
contradicts the modelling results by Jenista et al [80] (see
section 5.3), where the total heat transfer value remained about
the same for the diffuse and the constricted modes, while the
heat flux distributions changed significantly. This discrepancy Figure 29. Heat flux probe for anode heat flux measurements [87].
may be due to different conditions assumed for the model.
Cronin [62] measured the anode heat transfer for the wall
heat conduction probe is to transfer the heat impinging on the
stabilized arc configuration with varying flow parallel to the
sensing surface through the probe without any heat exchange
anode, finding for no lateral flow a total heat flux ranging
with the surrounding anode, i.e. to create a one-dimensional
from 1.3 to 2 kW for a 100 A argon arc, depending on the
heat flow field in the trunk of the probe. By measuring the
amount of flow directed towards the anode. Similar values are
temperature at two different locations within the trunk of the
reported by Iwao et al [61]. It has also been reported that an probe using thermocouples, the heat flux can be calculated with
increase in lateral gas flow lead to an attachment constriction the known thermal conductivity of the material (Cu) making
with an increase in arc voltage, and that the increased energy up the trunk of the probe.
dissipation lead to an increase in anode heat transfer. As an example, figure 30 [87] shows measured heat fluxes
From the measured heat transfer results one may at the anode of a free-burning argon arc (I = 200 A) at
conclude that p = 100 kPa, including a comparison with results from Nestor
(1) the energy transport associated with the electron flux is [81] obtained with a split anode arrangement and referring to
the dominating term in the anode heat transfer and an arc gap of 0.63 cm with the other parameters identical. The
(2) the change in total heat transfer when the attachment mode results of Nestor fall, as expected, between data for a gap of
is changed depends on the overall fluid dynamics more 1.0 and 0.5 cm. As the gap increases beyond 0.5 cm, the arc
than on the flux density distribution, i.e. it is important tends to be non-symmetric [87]. The heat fluxes presented in
how the mode change is accomplished, e.g. to constriction figure 30 are total heat fluxes composed of contributions due to
by reducing the flow directed towards the anode or by electron condensation at the anode, enthalpy transport by the
cooling the arc through lateral flow. electrons, heat conduction from the plasma to the anode, etc.
The Thomson scattering data of electron temperature and
What concerns the measurements of heat flux distribution, density fields previously described have been used to derive
for many years, a widely accepted approach was the split the heat fluxes associated with electron fluxes, i.e. electron
anode arrangement [81–86]. By splitting the anode in half and condensation, electron enthalpy and electron conduction (see
putting a thin (0.1 mm) electrically and thermally insulating section 5.3) [88]. The heat flux distribution over the anode
material between the two halves, the heat flow to each half surface are shown in figure 31 [88] for four different anode
can be measured calorimetrically. Measurements were made attachment modes: a diffuse mode, a lift-off mode, a multiple
by moving the split anode relative to the axis of a rotationally spot mode and a constricted mode.
symmetric arc. From the heat flow to each half of the anode as It is evident that for all attachment modes of the 100 A
a function of arc position, the heat flux profile may be obtained argon arc the electron condensation term is dominant, and
using the Abel inversion. that the peak heat flux increases from about 900 W cm−2 in
A later approach made use of a miniature heat conduction the diffuse mode, to about 1200 W cm−2 in the lift-off mode
probe shown schematically in figure 29 [87]. The probe area indicating higher current densities at the centre of the arc,
exposed to the arc is less than 1 mm in diameter, i.e. it provides to about 4 kW cm−2 in the constricted mode. Numerical
a high spatial resolution. The underlying principle of this integration of the experimental distributions for the rotationally

18
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

5.1. Flow perpendicular to the anode


As previously mentioned, the shape of the anode arc root
depends strongly on the energy balance in the anode boundary
layer. For example, energy transport by convection into the
anode boundary layer will encourage a diffuse anode arc
root. In contrast, if the arc has to generate sufficient energy
in the anode boundary layer to ensure the proper electrical
conductivity for passage of the current, the resulting arc root
will be more constricted.
The principal assumptions and results of the sheath model
by Dinulescu and Pfender [9] have been described in section 2.
Further developments have been pursued by Lowke and
his co-workers, in particular with improving the calculation
methodology [89–91]. In their approach, electron and ion
densities are calculated from balance equations including
ambipolar and regular diffusion. A constant electric field is
assumed when integrating from the arc column towards the
anode surface until the difference between electron and ion
densities exceeds a specified value. This way the difficulty is
avoided of solving Poisson’s equation for small space charges.
The electron density is assumed to be zero at the anode
Figure 30. Anode heat flux distributions for two different anode surface, and constant values are assumed for the mobilities
gaps [87]; the middle distribution was measured by Nestor with a and electrical conductivity throughout the boundary layer. The
split anode arrangement [81]. results for a 200 A free-burning argon arc show a field reversal
approximately 0.1 mm in front of the anode and an anode fall
symmetric diffuse mode gave total heat fluxes due to electron of approximately −2 V. In this approach, the effect of mass
transport of 735 W, which is about 50% of the total measured transport and, therefore, differences between constricted and
heat flux, in agreement with the earlier results by Sanders and diffuse attachments could not be described. A modification of
Pfender [13]. this sheath model has been used in the unified two-dimensional
description of an arc including cathode and anode [90]. Based
5. Modelling results for the anode region on the results of the earlier study, space charge effects have
been neglected in the calculation of the current and energy
The anode region, as any other part of an electrical discharge, transfer for a free-burning 200 A argon arc, and an effective
is governed by the conservation equations including the electrical conductivity including the diffusion terms is used.
current conservation and Maxwell’s equations. Unfortunately, The results for this configuration agree well with experiments
any attempt to solve these equations for the anode region justifying the assumptions, but no information on the potential
faces several major problems. First of all, the conventional distribution in the sheath region is obtained, and only a diffuse
conservation equations apply only as long as the continuum attachment is described. Tanaka et al [38] attempt to describe
approach is valid. Since the extent of the direct anode fall the conditions for different attachment forms by specifying
is in the order of one or a few mean free path lengths of the current density values for a free-burning argon arc at a location
electrons, the continuum approach is no longer valid for that 1 mm in front of the anode. By specifying a higher current
part of the anode region. Secondly, the application of the density for a 150 A arc than for a 50 A arc they obtain the result
conservation equations is more difficult because the plasma that the 150 A arc is constricted while the 50 A arc is diffuse,
is not in LTE and non-equilibrium properties are required. contrary to experimental observations with free-burning arcs.
Furthermore, the specification of realistic boundary conditions The model by Lowke et al [90] has been used for describing
at the anode surface requires assumptions which in the end the arc–anode interaction for the case of a molten metal anode
have a strong effect on the calculated results. In addition, a of a free-burning argon arc. The heat transfer and the drag on
numerical problem is encountered in calculating the potential the surface are calculated as well as the various forces leading
distribution in the sheath because solution of the Poisson to mass transport in the molten metal pool, including the effect
equation together with the conservation equations is difficult of varying viscosity and surface tension. Heat transfer to
for small differences between electron and ion densities. Most the anode by electron enthalpy flux and by thermodiffusion
modelling approaches use a one-dimensional sheath model has been neglected. In this model, a negative anode fall is
together with a two- or three-dimensional model for the arc. implicitly assumed because the dominating electron diffusion
The sheath model can therefore be applied to any arcing fluxes in the boundary layer require the field reversal [91].
configuration. However, since most calculations have been A recent model by Lago et al [92] assumes LTE throughout
performed for free-burning arcs, the sheath model is discussed the boundary layer and gets around the problem of current
in the section describing modelling approaches for the arc with transport through a cold gas layer by assuming an electrical
flow perpendicular to the anode. conductivity of copper in the boundary layer, but the other

19
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 31. Radial distributions for the different components of electron heat flux to the anode for (a) a diffuse attachment, (b) a lift-off
attachment, (c) a multiple spot attachment and (d) a constricted attachment [88].

properties remain that of the gas. A positive anode fall of 3.5 V is acceptable considering the assumptions made for the
is assumed for a free-burning 200 A argon arc. The boundary calculations. A further development of a similar model
condition for the electron density at the anode surface has been included the introduction of electron density non-equilibrium,
addressed by Cappelli and by Meeks and Cappelli [20, 21], as i.e. calculation of electron densities with balance equations
well as the effects of fluid flow on the attachment in a stagnation including diffusion, and space charge sheaths calculated from
flow model. Electron recombination at the anode surface at the Poisson equation [32, 80]. Both configurations have been
various rates is assumed as the dominant electron loss term, considered, that of a free-burning arc with the results of
and a strong effect on the electric field near the surface is the model by Hsu and Pfender [94] providing the boundary
demonstrated. While the assumptions for this model are more condition at the arc column, and that of a wall constricted
consistent with those for a low current arc (e.g. free stream arc with a gap in front of the anode. Different mass flow
temperatures of 4000 K), it is concluded that the convective rates towards the anode have been assumed to demonstrate
ion flux into the boundary layer must compensate for the ion the effect of the fluid dynamics. This model has been used
losses by diffusion and recombination, or a constriction and to demonstrate the effect of fluid dynamics by Amakawa et al
increased Joule heating will be needed for ion production. [35]. The following results taken from these models illustrate
The constriction of the arc in front of the anode has the physical processes in the anode boundary layer.
been modelled with an equilibrium approach and with a two- Figure 32 [32] shows that the electron temperatures start
temperature non-equilibrium approach by Chen and Pfender deviating from the heavy particle temperatures about 0.2 mm
[10, 93], and the ‘thermal pinch’ is demonstrated by these in front of the anode surface for a diffuse attachment, and
modelling results for the configuration of a wall constricted that they remain above a value of 10 000 K up to the anode
arc with a 10 mm gap between the constrictor and the surface. Figure 33 shows the calculated streamlines for both
anode. However, the details of the processes in the sheath the diffuse and the constricted mode for a 200 A argon arc [35].
are not described in these models. The results of this With a superimposed mass flow rate of 0.2 g s−1 , the plasma
model are compared with experimental results in figure 9 emanating from the confinement tube impinges on the anode
[14]. The agreement between calculations and measurements and is forced to flow out in the radial direction. A relatively

20
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

in figure 34 for both the constricted and the diffuse mode.


On the other hand, if the potential curves are extrapolated
from the arc column to the anode with a constant slope, and
the anode fall is defined as the potential difference between the
intersection of the extrapolated potential curve with the anode
and the anode potential (see figure 2), we obtain for the diffuse
mode (curve A) a negative fall, whereas the anode fall for
the constricted mode (curve B) assumes now positive values.
In the latter case, the anode fall includes the arc constriction
region, i.e. it extends over more than 5 mm from the anode
surface. One of the main distinctions between the diffuse
and the constricted modes is the current density at the anode
which is substantially smaller for the diffuse mode. The higher
current density leads to an increase in the axis temperature in
the constriction region.
Figure 32. Electron and heavy species temperature distributions in
Calculations of the centreline profiles of the electrical
front of the anode (right border) of a 200 A argon arc. (Reprinted
from [32] copyright 1997 IEEE.) potential in the anode region for three different currents
are shown in figure 35 [32]. This figure refers to a
small amount of gas from the surroundings is entrained into partially confined argon arc at atmospheric pressure with a
the arc fringes (figure 33(a)). By drastically reducing the superimposed flow of 0.165 g s−1 for all three cases. The
superimposed mass flow rate (0.01 g s−1 ), the arc constricts potential curves for the different currents are similar showing
in front of the anode and an anode jet is formed. a reversal of the electric field in front of the anode and
The latter leads to a strong entrainment of gas from the corresponding negative voltage drops ranging from 0.4 V
surroundings and this gas flow opposes the superimposed (50 A) to 1.3 V (300 A).
flow (figure 33(b)), forming a stagnation point close to the As previously mentioned, the steep gradients in the anode
end of the confinement tube (approximately 9 mm above boundary layer may make substantial contributions to, or even
the anode surface for a 10 mm spacing between the end of dominate, the current flow as described by equation (2). The
the confinement tube and the anode surface), whereas the various terms in this equation have been calculated for the
stagnation point in figure 33(a) is essentially at the anode [35]. anode boundary layer of a free-burning argon arc at 200 A and
Referring to the same configurations as in figure 32, the atmospheric pressure [32].
potential distributions on the axis of the arc between the Figure 36 [32] shows the results for the axis of the arc
end of the confinement tube and the anode surface (gap of over a distance of 5 mm from the anode (the location of
10 mm) have been calculated, considering a 200 A arc in the anode is at 5 mm in figure 36). The applied electric
argon at atmospheric pressure [32]. With the anode potential field is primarily responsible for the current flow at distances
as reference (V = 0), figure 34 [32] shows the potential >1 mm from the anode. At a distance of approximately
distributions across this gap for the diffuse mode (curve A) 0.7 mm from the anode, the electric field changes sign which
with a superimposed mass flow of 0.165 g s−1 and for the is necessary to ensure current continuity, i.e. this negative
constricted mode (curve B) with the superimposed mass flow field will repel some of the electrons driven by diffusion
rate reduced to 0.008 g s−1 . The constricted mode shows towards the anode. The actual flow of electron current
a substantially higher overall voltage drop between the end to the anode is obtained by subtracting the reverse current
of the constrictor and the anode (∼5 V) compared with the (curve 1) from the diffusion current (curve 2) and adding
diffuse mode (∼2.4 V). This higher voltage drop is due to arc the current due to thermal diffusion. According to figure 36,
constriction and the associated entrainment of cold gas which this current density is ∼450 A cm−2 including the contribution
extends to more than 5 mm from the anode [32]. due to thermal diffusion (∼90 A cm−2 ). As these numbers
At a distance of ∼2 mm from the anode, the plasma show, thermal diffusion needs to be taken into account. Since
potential starts to exceed the anode potential, indicating that there is also a positive ion current driven to the anode, the
a positive space charge is being formed. This leads to an net current density is obtained by including this ion current.
electric field reversal which occurs ∼0.5 mm from the anode By extrapolating experimental results discussed in the current
surface (curve B), i.e. there is a negative voltage drop in front probe measurements, the ion current density at the anode will
of the anode of ∼1.4 V, a result which had not been expected. be ∼60 A cm−2 for these conditions, i.e. the net current density
This result points to the difficulty of defining and measuring at the anode in the axis of the arc will be ∼390 A cm−2 .
anode falls, because the voltage drop in front of the anode This value is somewhat higher than the value derived from
does not only depend on the location of the measurement measurements of a wall stabilized arc, but lower than the values
point, but also on the mode of the anode arc attachment which obtained from probe measurements on free-burning arcs.
is determined by the flow. If the anode fall was defined as Ion current densities for the anode region have been
the potential drop immediately in front of the anode (over calculated for the same configuration and the same operating
a distance 0.1 mm from the anode which is essentially the conditions applied in figure 36. The results are shown in
thickness of the sheath), the anode fall is negative as seen figure 37 [32]. Similar to the case of a free-burning arc,

21
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 33. Stream lines in the anode region of a 200 A argon arc with a diffuse (a) and a constricted (b) attachment; the anode surface is the
lower boundary, on the upper right hand side. (Reprinted from [35] copyright 1998 IOP Publishing.)

Figure 36. Contributions to the electron current in the anode


Figure 34. Potential distribution in front of the anode of a 200 A boundary layer of a 200 A argon arc; the anode surface is at the right
argon arc; the anode is on the right hand border. (Reprinted hand boundary. (Reprinted from [32] copyright 1997 IEEE.)
from [32] copyright 1997 IEEE.)
of the current [48], whereas in the wall stabilized arc there is
a constant superimposed flow. It is interesting to notice that
the ion current densities show a maximum in front of the anode
which is associated with the space charge distribution and the
behaviour of the electric field (figure 36) in front of the anode.
A special case has been modelled by Nemchinsky [95],
where an anode is considered with a diameter smaller than
the arc diameter (corresponding to the situation in gas metal
arc welding). The one-dimensional model assuming LTE
properties predicts increased voltage drops in the anode
boundary layer for decreasing anode diameters and increasing
current densities. A maximum is shown for the electron
temperature in front of the anode. These results are consistent
with those assuming a constricted attachment due to fluid
dynamic conditions and are similar to those obtained in HID
Figure 35. Potential distribution in front of the anode of an argon lamp arcs [96].
arc for different arc currents; the anode surface is on the right hand Benilov and his co-workers have recently applied a
border. (Reprinted from [32] copyright 1997 IEEE.) model originally developed for the cathode region also to the
non-equilibrium part of the anode boundary layer [97, 98].
the ion current densities increase with increasing current, but The model was applied to high pressure (100 MPa) Xe and
their magnitudes are different which is probably due to the Hg arcs in configurations and with operating parameters
different flow situation. In the case of the free-burning arc, similar to those encountered in high pressure arc lamps.
there is only the self-induced flow which is a strong function The model describes the region from the anode surface to

22
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

simultaneously with appropriate boundary conditions. The


following results refer to the operating conditions:
arc current : I = 100 A,
gap between constrictor tube and anode: d = 5 mm,
axial flow superimposed to the arc: Qarc = 5 slpm,
cross flow (cold flow): Qcold =18.0 slpm.
Figure 38 [101] shows the distribution of the heavy particle
temperature (a), of the electron temperature (b), of the number
density of electrons (c) and of the current density (d). All these
diagrams refer to the symmetry plane, i.e. the plane passing
through the axis of the constrictor tube, perpendicular to the
frontal plane of the cross flow injector. The mass flow rate
of the cross flow is kept relatively small in order to avoid arc
instabilities (restrike).
Figure 38(a) shows the isotherms of the heavy particles
Figure 37. Ion current distribution in front of the anode of different and the shape of the deflected arc. There is a drop in the heavy
arc currents; the anode surface is at the right hand border. particle temperature in front of the anode over a boundary layer
(Reprinted from [32] copyright 1997 IEEE.)
of thickness in the order of 1 mm. As a consequence, one
would expect the arc root to be somewhat constricted rather
150 µm in front of it, and a self-consistent model for the than diffuse. This seems to be corroborated by figures 38(b)
diffusion fluxes is used [99]. The current density is given and (c) which show that high values of electron temperature and
as an independent parameter, simulating different degrees electron densities in front of the anode are confined to rather
of constriction; however, the values assumed are quite high small areas, consistent with the current density distribution at
with 107 –108 A m−2 , significantly higher than those observed the anode (figure 38(d)). By defining the arc deflection as the
in low current arcs at 0.1–1 MPa, and the higher value difference of the current density maximum without and with
exceeding those usually observed in constricted high current cross flow, the distance amounts in this case to ∼6.7 mm [101]
attachments. Electron and heavy species temperature and in good agreement with experimental observations [61].
density distributions are obtained as well as the potential Recently, several three-dimensional models have been
distribution. The results indicate a potential drop immediately developed describing the dynamics of an arc in a plasma torch
in front of the anode and a small peak in electron temperature with moving anode attachment [102–107]. These models
values, but closer to the anode surface than those observed at describe the interaction of the arc with the axial flow quite
atmospheric pressures. Since these results have more relevance realistically; however, they do not describe the details of the
to low current arcs, they are discussed more in the review on anode attachment. Rather artificial assumptions are made
HID lamp anodes [96]. to allow current transfer to the anode through the cold gas
boundary layer. These models are not discussed in this review.
5.2. Flow parallel to the anode
Flow parallel to the anode is encountered in many arc 5.3. Anode heat flux
plasma torches, where the axis of the anodic arc is initially The total heat flux to the anode is described by
perpendicular to the anode and to the flow (see figure 15).
dTe dTh
This configuration has been used for recent experimental qa = je φa + qe − ke − kh + ji (Ei − φa ) + QR , (5)
studies [61, 63, 64, 100] of arc/flow interactions and arc dz dz
instabilities and these studies also revealed a relatively large where je and ji are the electron and ion current densities, φa
non-equilibrium region of the deflected arc in front of the the work function of the anode material, Te and Th and ke and
anode with relatively high electron densities and electron kh are the electron and heavy particle temperatures and thermal
temperatures, substantially higher than the equilibrium values. conductivities, respectively, Ei is the ionization potential of the
This type of plasma environment would be advantageous for gas atom and QR is the radiated energy transfer. qe describes
plasma processing of materials. the transport of energy associated with the electron flux. The
For modelling of the anode region of the arc exposed first term on the right-hand side describes the ‘condensation’
to cross flow, it has been assumed that the plasma flow is energy, i.e. the energy transferred to the anode when electrons
three-dimensional (3D), quasi-steady and laminar. A set of carrying the current are incorporated into the lattice of the
3D governing equations, similar to those described in [43], anode material. The third and fourth terms are the heat fluxes
has been established, including the species mass conservation due to the temperature gradients in the electron and heavy
equation, the mass-averaged momentum equations, the particle gases. The fifth term describes the energy deposited
electron and heavy particle energy conservation equations, the when ions recombine at the surface; this term is usually small
potential equation and the magnetic vector potential equations. because the ion current is usually a small fraction of the electron
This set of partial differential equations has been solved current. In most cases, the first two terms and the heavy

23
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 38. Distributions of (a) heavy particle temperatures, (b) electron temperatures, (c) electron densities and (d) current densities in
front of the anode of a partially constricted arc with cold gas flow parallel to the anode; argon, 1 bar, 100 A. (Reprinted from [101] copyright
2005 IEEE.)

particle conduction are considered the most important ones. where φ is the thermal diffusion coefficient and σ the electrical
The electron energy flux term contains the thermal energy of conductivity.
the electrons, and the energy gain or loss in the sheath region: What concerns the two conduction terms, the electron
  conduction accounts for energy transfer by electrons not
kTe
qe = je 2 + Us (6) incorporated into the metal lattice, i.e. reflected from the
e surface. This term is usually very small. The heavy particle
with Us the voltage drop across the electrical boundary conduction assumes that the atoms and ions equilibrate with
layer (see figure 2). Two different assumptions have been the anode surface. It is not certain that this assumption can
made to calculate this term: (a) the electrons continue to be be made because in the case of a constricted attachment the
thermalized in the electrical boundary layer and an electron temperature gradients are so steep that significant changes can
temperature according to a MB distribution determines the be encountered over one mean free path length, resulting in
electron energy; in this case Us is zero; (b) the electron temperature slip conditions. In this case, the conduction term
energy is composed of a thermal component according to may be replaced by
the temperature at the sheath edge and a non-Maxwellian α(Th − Ta ) (8)
component obtained through acceleration/deceleration by the
with Th and Ta the heavy particle temperature in the sheath
sheath voltage drop; in this case Us is either positive or
and the anode temperature, respectively, and α the energy
negative). Boundary layer calculations using a continuum
accommodation coefficient for non-continuum slip conditions.
approach indicate that Us should be negative even for a
Because there will be a positive ion current ji , driven to
constricted attachment because of the high diffusion fluxes
[32], but the peak of the potential in front of the anode is the anode by the steep ion density gradients and a negative
in the order of 0.1–0.5 mm, i.e. many mean free paths, and anode fall, the energy transport due to this flux needs to be
thermalization of the energy gain should be assumed. On considered. Upon impact on the anode, the positive ions will
the other hand, potential distributions derived from Thomson neutralize, releasing an energy
scattering measurements neglecting the ion current indicate an
ji (Ei − φa ), (9)
increase in the potential to a position 50 µm from the anode.
In this case, it is unknown what sheath potential should be
where Ei is the ionization potential of the neutral atoms. In the
assumed.
case of a negative anode fall and temperature slip of the heavy
In the case of strong electron temperature gradients, a
species, the acceleration of the ions in the sheath region needs
thermodiffusion term should be added:
to be considered, and equation (9) needs to be replaced by
φTe
je , (7)
σ ji (Ei − φa + Us ). (10)

24
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

The last term in equation (5) accounts for radiation from the (4) in the case of the diffuse attachment there is a strong
plasma transferred to the anode, assuming that the plasma is convective transport from the arc column into the
optically thin. anode region; however, this convective transport is
In summary, one can use for the calculation of the heat distributed over a wider area, and each individual term in
flux to the anode equation (12) is smaller than in the constricted attachment
 
kTe φTe case. However, the total heat flux in the case of the diffuse
qa = je φa + je 2 + Us + je + α(Th − Ta ) attachment may be comparable or even larger than in the
e σ
case of the constricted attachment;
+ ji (Ei − φa ) + QR (11) (5) experiments as well as simulations indicate that for a
in the case of a positive sheath potential drop (in which case diffuse attachment about 50% of the total heat transfer
the ion flux term is very small), or to the anode is due to the electron flux, and this fraction
  is higher in the case of a constricted attachment.
kTe φTe
qa = je φa + je 2 + je + α(Th − Ta )
e σ Figure 39 [32] shows the contributions of the various
terms in the anode heat flux equation (equation (12)) to the
+ ji (Ei − φa + Us ) + QR (12)
anode heat flux. The electron enthalpy makes a significant
in the case of a negative sheath potential drop. contribution to the total anode heat fluxes as shown in figure 39
The difference between the constricted and the diffuse for a 100 A argon arc operated in the cathode jet dominated
attachment can be described by considering an integral energy mode with a superimposed flow of ṁ = 0.165 g s−1 [32].
balance of the boundary layer: The largest contribution, however, is by electron condensation,
ṁh − Qa − Qcon − Qrad + I (Ua + Ecol z) = 0 (13) followed by heavy particle conduction. The reason for this
large contribution is, to a large degree, due to the superimposed
with ṁ the plasma gas mass flow rate into the boundary layer, flow. The contributions of electron conduction and ion
h the change in gas enthalpy from entering and leaving the recombination are almost negligible. The contribution of
boundary layer, Qa the energy loss to the anode, Qcon the net radiation is not included in this diagram.
conduction loss (axial conduction from the column and radial Figure 40 [80] shows the radial distributions of the total
loss), Qrad the radiative loss (assuming optically thin radiation) heat flux to the anode for a 200 A argon arc considering both
and the last term indicating the energy dissipation between the a diffuse and a constricted arc root. Although the peak heat
column end and the anode surface, with z the boundary layer flux is almost four times higher for the constricted arc root,
thickness and Ecol the potential gradient in the column (see the integrated total heat flow is almost the same for both cases.
figure 2). Integrity of the anode, however, is determined by the peak heat
In the case of a constricted attachment, an anode jet fluxes rather than the integrated values. As recently shown by
directed away from the anode surface will carry energy towards Terasaki et al [108] melting of the anode in a free-burning arc
the column, and the first term in equation (13) is negative. The is determined by the peak heat flux rather than the total anode
other loss terms are increasing to some degree. This energy loss heat transfer.
is compensated by increased dissipation, i.e. a higher electric Figure 41 shows the radial distribution of the total anode
field in the constriction region (see figure 2). This increased heat flux for a free-burning 200 A argon arc at p = 100 kPa
electric field results in a voltage drop in the constriction region and an electrode gap of 1.0 cm [80, 87]. Also included in
which is usually larger than the relatively small voltage drop this diagram is a comparison with experimental data [87]. As
immediately in front of the surface which results from the previously mentioned, the arc tends to be non-symmetric for
increased charge transport by diffusion. The total anode fall, arc gaps exceeding 0.5 cm.
considered to be the sum of the voltage drop in the constriction For this reason, figure 41 shows both the left and
region and in the sheath region, is then positive. The increased the right side of the measured radial heat flux distribution.
energy dissipation in the boundary layer is transferred to the This, however, limits comparisons with analytical predictions.
anode, as verified by experiments [61]. However, for a diffuse attachment of a 100 A argon arc,
Comparing the different energy transfer terms for the electron heat flux data derived from Thomson scattering
different attachment modes, we find the following: experiments show good agreement with the theoretical
(1) energy transfer due to electron condensation is the same predictions as shown in figure 42 [88]. In these results, the
in both cases; mass flow towards the anode was assumed to be 0.165 g s−1 in
(2) energy transfer due to electron enthalpy flux may be the model and was 0.446 g s−1 in the experiment. The total heat
slightly higher in the case of a constricted attachment transfer to the anode was modelled to be 1310 W and measured
because of higher electron temperatures; as 1500 W, with the electron flux contribution 735 W [88].
(3) an additional energy transfer exists in the case of
the constricted attachment due to the increased energy 6. Comparison of the anode regions of low and high
dissipation in the anode region; this increased energy flux intensity arcs
may be associated with an increase in the conduction terms
because of steeper temperature gradients, with an increase A reduction in arc current usually results in a change in the
in the enthalpy flux and with an increase in the energy arc attachment mode to a constricted attachment, as described
transport associated with the electron current; by several investigators, e.g. [30, 69], the change occurring

25
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Figure 39. Contributions to the heat flux to the anode surface of a Figure 41. Comparison of calculated and experimental anode heat
100 A argon arc with a diffuse anode attachment. (Reprinted flux distributions; calculated fluxes are from [32], measured data
from [32] copyright 1997 IEEE.) from [87]. The experimental data show an asymmetric attachment,
and measurements for both sides are shown for comparison.

Figure 40. Anode heat flux distribution for a 200 A argon arc with a
diffuse and a constricted attachment. (Reprinted from [32] Figure 42. Radial distributions of the different components for the
copyright 1997 IEEE.) anode heat flux, calculated (lines) [32] and derived from Thomson
scattering measurements (symbols) [88].
between 20 and 40 A. This change is likely caused by the
need for increased ionization processes to compensate for the
loss of energy and of electrons to the anode, i.e. increased The following observations can be made:
energy dissipation. The studies by Mentel and Heberlein (1) All low current arcs have an electron temperature
[96] have concentrated on a special case of low current arcs maximum between the column and the anode. This is
operated between rod shaped tungsten electrodes where the in part due to the fact that the anode diameter is smaller
anode is sufficiently hot to thermionically emit electrons. This
than the arc diameter, resulting in a constricted attachment
fact probably contributes to the observed peak in electron
which may show a moderate constriction and relatively
densities in front of the anode surface in these experiments
low current densities. The cooling of the plasma in front
[96, 109] even though the current densities are similar to those
of the anode at higher anode temperatures can be the
observed in diffuse attachments of high intensity arcs. Table 4
consequence of interactions of the plasma with electrons
presents a comparison of electron density and temperature
which are thermionically emitted and reabsorbed by the
measurements at different distances from the anode for a low
current arc and a high intensity arc operated under different anode. However, this electron temperature maximum is
conditions. also apparent for low anode temperatures for low current
The data in table 4 were obtained with a model arc and for constricted high intensity arcs. This is in contrast
lamp using emission spectroscopy for the low current arcs to high intensity arcs with a diffuse attachment where
[96], and with a wall stabilized arc with a gap in front of there is a monotonic drop in the electron temperature and
the anode using Thomson scattering for the high intensity density values towards the anode.
arcs [66]. The attachment mode in the high intensity arcs (2) The electron temperature values at the arc column increase
was controlled by adjusting the axial gas flow rate towards only by about 30% for an increase in the current from 5
the anode. Measurements by other authors show similar to 120 A, whereas the electron density values increase by
values [30, 65]. more than an order of magnitude.

26
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

Table 4. Comparison of plasma conditions in front of the anode for high intensity arcs, but estimates indicate that they can be
low current and high intensity arcs and different attachment extremely high, exceeding 109 W m−2 , and comparisons with
modes [66, 96]. The low current arc is operated with a rod shaped
low intensity arcs are no longer meaningful.
tungsten anode of 1.5 mm diameter and 20 mm length.
There are some differences in the approaches for the model
Current Position Te ne formulation for the anode attachment. The major ones are the
(A) from anode (kK) (m−3 ) Condition following:
1 1 mm, column 8 1.45 × 10 21
Constricted
(1) In the low current arcs, the electron flux can be described
1 150 µm, peak Te 12 Constricted
1 50 µm 11.5 4.5 × 10 21
Constricted by an electron saturation current at the sheath edge
5 >1 mm, column 8.5 3 × 1021 Constricted modified by the negative potential drop in the sheath,
5 800 µm, peak Te 11.4 Constricted which is assumed to be always present. In the high
5 400 µm, peak ne 4.3 × 1021 Constricted current arcs, the electron flux is determined by diffusion
5 50 µm 8.5 3.8 × 1021 Constricted (primarily) with an additional drift in the case of a positive
60 2.5 mm, column 11.2 2 × 1022 Diffuse potential drop.
60 50 µm 9 8.5 × 1021 Diffuse (2) The anode fall in the low current arcs is always positive due
120 2.5 mm, column 13.5 4.7 × 1022 Diffuse
120 50 µm 12.4 1.8 × 1022 Diffuse to an increase in the electric field in a constriction region.
100 1.8 mm, ne peak 14.4 6.2 × 1022 Constricted The sheath voltage drop seems to be always negative,
100 50 µm 14.4 3.9 × 1022 Constricted but the magnitude is smaller than the initial increase in
100 50 µm 10.8 1.5–3.3 × 1022 Diffuse the potential. In high intensity arcs, the reported anode
fall values are inconsistent in part because the reference
point at the column end is usually not known. This is
(3) For a constricted attachment there is only a small or even particularly true for negative falls where the potential
no temperature drop between the peak temperature and the gradient shows a monotonic decrease going from the
value at 50 µm for high intensity arcs. The temperature column towards the anode. Considering an electric field
values for high current arcs (100 versus 1 A) are only about in the arc column of 5–10 V cm−1 , this could add an
25% higher, whereas the electron density values at 50 µm uncertainty of 0.5–1 V to the reported data.
from the anode are almost an order of magnitude higher (3) There are some minor differences in the treatment of the
than those in front of low current arc anodes. energy fluxes. The heavy particle energy transfer to the
(4) For a constricted attachment there is a significant electron anode is described in separate terms for the ions and for
density peak between the column and the anode for the the neutrals in the treatment by Mentel and co-workers
high intensity arc. Such an electron density peak in the [96, 109], and the electron conduction term is neglected.
boundary layer is also seen for currents in the 1.5–5 A The power dissipation in the boundary layer is given for
range with hot anodes; however, for the 1 A arc the the low current arc as the product of arc current and
electron densities increase monotonically to the position the potential difference between the column end and the
of 50 µm. anode, and is always positive. For the diffuse attachment
There are also similarities in the current density and power in the high intensity arcs with a monotonically decreasing
flux data. Average current densities for low current arcs potential gradient, this dissipation term can be negative
are between 60 and 130 A cm−2 for currents between 1 and because there is strong convective energy transport into
5 A, compared with current density values between 100 and the boundary layer.
200 A cm−2 for diffuse attachments with currents between 100 The electron condensation term I a is found to be the strongest
and 200 A. It should be noted that these low current arcs contributor to the heat flux in high intensity and low current
are called constricted even though their current densities are arcs. The work function a and with it the heating of the anode
comparable to those of diffuse high intensity arcs because may be reduced by a monolayer of special emitter material,
the power dissipation in the anode boundary layer is largely e.g. of alkali metals. An example is the reduced heating of
transported to the arc column and away from the anode. This the tungsten anode in a high pressure sodium lamp [109]. But
energy transfer is mainly due to convection which results no difference has been found in the heating of anodes made of
from magnetic forces produced according to Maecker by pure and thoriated tungsten in carefully controlled experiments
the increase in current density towards the anode. But in the model lamp [96, 109]. One could hypothesize that
the constricted attachment for high current arcs has current only a small part of the anode surface is covered with a
densities exceeding 500 A cm−2 . Values of several kA cm−2 thorium monolayer so that the average work function remains
are possible for currents of 100–300 A. These values typically rather unchanged. Of interest should be to note that different
lead to anode material evaporation and therefore cannot be materials do have an effect on the plasma in front of the anode
used for a valid comparison. for high intensity arcs as shown by Gleizes et al [33].
The power flux densities are about 1 kW cm−2 for a 3 A Finally, it needs to be emphasized that the self-magnetic
arc, and about 1.5 kW cm−2 in the case of a diffuse attachment pressure increases with the product of current and current
of a 100 A arc, with the total power transferred to the anode density, and fluid dynamic effects induced by magnetic
being 18 W for the 3 A arc and 1500 W for the 100 A arc. There pressure gradients are much stronger with high intensity
are few if any power density measurements for constricted arcs. The consequence is that the transition to a constricted

27
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

attachment mode and the formation of an anode jet appears as the authors also acknowledges the Alexander v Humboldt
an instability, i.e. as an abrupt transition in the case of high Foundation for granting the Senior Scientist Award during
intensity arcs. which this survey was initiated.
The authors acknowledge the permission to reproduce
7. Conclusions some figures from the following copy right holders: Institute
of Physics, figures 10, 11, 12, 15, 17, 18, 19, 20, 21, 27, 28 and
The arc anode attachment of high intensity arcs is strongly 33; American Institute of Physics, figures 3, 4, 5 and 9; IEEE,
affected by the fluid dynamics of the arcing arrangement. figures 32, 34, 35, 36, 37, 38, 39, 40 and 41; ASM International,
Current and heat flux densities can change by more than a figure 13: Université de Rennes, France, figure 31; Université
factor of four for almost identical conditions if the attachment de Orleans, France, figure 16; Dr E Leveroni-Calvi,
changes from a diffuse to a constricted one. While results figures 22, 23, 24, 25 and 26; Dr J Menard, figures 29
of electron temperature and density measurements are in and 30.
general in agreement in different studies, results of anode fall
measurements differ quite widely. One of the major reasons
is the difficulty of defining a reference point from which to References
measure the potential change. As a consequence, positive [1] Finkelnburg W and Maecker H 1956 Encyclopedia of Physics
and negative values for anode falls have been reported for ed S Flügge vol XXII (Berlin: Springer) p 254
similar conditions. However, considering theoretical as well [2] Ecker G 1961 Naturwissenschaften 33 1
as experimental results, one may conclude that for diffuse [3] Guile A E 1971 IEE Proc.118 1131
attachments a potential maximum exists in front of the anode [4] Pfender E 1978 Electric arcs and arc gas heaters Gaseous
Electronics ed M N Hirsh and H J Oskam (New York:
with a field reversal to control the electron diffusion flux.
Academic) pp 291–398
If the reference point is the potential extrapolated from the [5] Mitchner M and Kruger C H J 1973 Partially Ionized Gases
column to the anode position with a constant potential gradient, (New York: Wiley)
the anode fall is always negative for such an attachment. If the [6] Dresvin S V (ed) 1977 Physics and Technology of
reference point is the potential at the end of the arc column, the Low-temperature Plasmas (Ames, IA: The Iowa State
anode fall can be negative, zero or slightly positive depending University Press)
[7] Boulos, M I, Fauchais P and Pfender E 1994 Thermal Plasmas:
on the boundary layer thickness. However, the location of Fundamentals and Applications (New York: Plenum)
this reference point is difficult to determine. In a constricted [8] Fridman A and Kennedy L A 2004 Plasma Physics and
attachment, an increase in the potential gradient is present; Engineering (New York: Taylor and Francis)
however, it is at present uncertain whether over the last mean [9] Dinulescu H A and Pfender E 1980 Analysis of the anode
free paths a reversal of the gradient takes place, as has been boundary layer of high intensity arcs J. Appl. Phys.
51 3149–57
postulated by simulations using a continuum approach, or [10] Chen D M and Pfender E 1980 Modeling of the anode
whether the potential keeps increasing throughout the sheath. contraction region of high intensity arcs IEEE Trans.
Modelling of the anode region has relied on two- Plasma Sci. 8 252–9
dimensional CFD models coupled to a one-dimensional sheath [11] Chen D M and Pfender E 1981 Two-temperature modeling of
model. Recent experimental results have shown that (a) the anode contraction region of high-intensity arcs IEEE
Trans. Plasma Sci. 9 265–74
there are strong radial electron density gradients and diffusion [12] Sanders N, Etemadi K, Hsu K C and Pfender E 1982 Studies
fluxes resulting in rather complex current density distributions of the anode region of a high-intensity argon arc J. Appl.
and a one-dimensional approach may not be always justified. Phys. 53 4136–45
Furthermore, the steep gradients encountered in a constricted [13] Sanders N A and Pfender E 1984 Measurement of anode falls
attachment raise the question if a continuum approach can and anode heat transfer in atmospheric pressure high
intensity arcs J. Appl. Phys. 55 714–22
be pursued in a region where the plasma properties change [14] Hsu K C, Etemadi K and Pfender E 1983 Study of the
significantly over one mean free path length. free-burning high-intensity argon arc J. Appl. Phys.
Comparisons with low current arc anode attachments 54 1293–301
reveal surprising similarities with regard to plasma properties [15] Etemadi K and Pfender E 1985 Impact of anode evaporation
and current and heat flux densities. However, it must be on the anode region of a high-intensity argon arc Plasma
Chem. Plasma Process. 5 175–82
emphasized that these comparisons were made with a low [16] Golubovskii Y B, al Hawat S H and Tsendin L D 1987
current arc with a thermionically emitting anode, serving as Electron distribution in the anode region of a low-current
a source of electrons and thus avoiding strong constriction. discharge in neon Sov. Phys.—Tech. Phys. 32 760–3
Little information is available for strongly constricted [17] Zhukov M F, An’shakov A S, Vashchenko S P,
anode attachments with significant material evaporation, and Dandaron G N B and Zayatuyev K 1987 Formation of
contracted anode spot in generators of low-temperature
no results are presented in this review. The area of highly
plasma Sov. J. Appl. Phys. 1 71–6
constricted anode attachments is at present poorly understood. [18] Westhoff R and Szekely J 1991 A model of fluid, heat flow,
and electromagnetic phenomena in a nontransferred arc
Acknowledgment plasma torch J. Appl. Phys. 70 3455–66
[19] Ushio M, Sadek A A and Matsuda F 1991 Comparison of
temperature and work function measurements obtained with
The authors acknowledge support from the US National different GTA electrodes Plasma Chem. Plasma Process.
Science Foundation, Award No CTS-0225962. One of 11 81–101

28
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

[20] Cappelli M A 1993 The nonequilibrium region of an electrode [41] Gonzalez J J, Freton P and Gleizes A 2002 Comparisons
in contact with a flowing thermal plasma IEEE Trans. between two- and three-dimensional models: gas injection
Plasma Sci. 21 1–6 and arc attachement J. Phys. D: Appl. Phys. 35 3181–91
[21] Meeks E and Cappelli M A 1993 Two-temperature fluid model [42] Li H P and Chen X 2001 Three-dimensional modelling of a dc
for high-pressure plasmas in contact with cooled electrodes non-transferred arc plasma torch J. Phys. D: Appl. Phys.
J. Appl. Phys. 73 3172–82 34 L99–L102
[22] Lefort A, Parizet M J, El-Fassi S E and Abbaoui M 1993 [43] Li H P, Pfender E and Chen X 2003 Application of
Erosion of graphite electrodes J. Phys. D: Appl. Phys. Steenbeck’s minimum principle for three-dimensional
26 1239–43 modelling of dc arc plasma torches J. Phys. D: Appl. Phys.
[23] Sadek A A, Ushio M and Matsuda F 1990 Effect of rare earth 36 1084–96
metal oxide additions to tungsten electrodes Metall. Trans. [44] Nemchinsky V A and Peretts L N 1977 Anode sheath in a
A 21 3221–6 high-pressure, high-current arc Sov. Phys.—Tech. Phys. 22
[24] Ushio M, Fan D, and Tanaka M 1993 Contribution of arc 1083–7
plasma radiation energy to electrodes Trans. JWRI 22 [45] Bez W and Höcker K H 1954 Theorie des Anodenfalls I
201–7 Z. Naturf. a 9 72–81
[25] Razafinimanana M, Hamidi L E, Gleizes A, and Vacquié 1995 [46] Höcker K H and Bez W 1955 Theorie des Anodenfalls II
Experimental study of the influence of anode ablation on Z. Naturf. a 10 706–14
the characteristics of an argon transferred arc Plasma [47] Bez W and Höcker K H 1955 Theorie des Anodenfalls III
Sources Sci. Technol. 4 501–10 Z. Naturf. a 10 714–17
[26] Tanaka M, Tanaka K, and Ushio M 1996 Calculating the [48] Maecker H 1955 Plasmaströmungen in Lichtbögen infolge
temperature of electrodes in gas-tungsten-arcs Trans. JWRI eigenmagnetischer Kompression Z. Phys. 141 198–216
25 9–15 [49] Hermoch V 1963 The anode space of short-time high-intensity
[27] Ushio M, Ikeuchi K, Tanaka M and Tsuyoshi H 1996 Plasma electric discharges Czech. J. Phys. B 13 327–34
sintering of new carbide electrodes Trans. JWRI 25 17–23 [50] Hermoch V 1963 On the formation of the anode space of a
[28] Ushio M, Tanaka M and Wu C S 1996 Analytical approach to short-time high-intensity electric discharge Czech. J. Phys.
anode boundary layer of gas tungsten arcs Trans. JWRI 25 B 13 321–6
9–21 [51] Hartmann R M and Heberlein J V 2001 Quantitative
[29] Salihou H, Guillot J P, Abbaoui M and Lefort A 1996 Anode investigations on arc–anode attachments in transferred arcs
parameters of short arcs at low current J. Phys. D: Appl. J. Phys. D: Appl. Phys. 34 2972–8
Phys. 29 2915–21 [52] Wutzke S A, Pfender E and Eckert E R G 1967 Study of
[30] Dyuzhev G A, Mitrofanov N K and Shkol’nik S M 1997 electric-arc behavior with superimposed flow AIAA J. 5
Experimental investigation of the anode region of a 707–14
free-burning atmospheric-pressure inert-gas arc: I. General [53] Wutzke S A, Pfender E and Eckert E R G 1968 Symptomatic
characteristics of the discharge. Low-current regime Tech. behavior of an electric arc with a superimposed flow AIAA
Phys. 42 30–4 J. 6 1474–82
[31] Baksht F G, Dyuzhev G A, Mitrofanov N K and [54] Paik S, Huang P C, Heberlein J and Pfender E 1993
Shkol’nik S M 1997 Experimental investigation of the Determination of the arc-root position in a dc plasma torch
anode region of a free-burning atmospheric pressure inert Plasma Chem. Plasma Process. 13 379–97
gas arc: II. Intermediate current regime—multiple anode [55] Coudert J F, Planche M P and Fauchais P 1994 Anode–arc
constriction Tech. Phys. 42 30 attachment instabilities in a spray plasma torch High Temp.
[32] Jenista J, Heberlein J V R and Pfender E 1997 Numerical Chem. Processes 3 639–51
model of the anode region of high-current electric arcs [56] Duan Z, Wittmann K, Coudert J F, Heberlein J and Fauchais P
IEEE Trans. Plasma Sci. 25 883–90 1999 Effects of the cold gas boundary layer on arc
[33] Gleizes A, Bouaziz M, Gonzalez J J and Razafinimanana M fluctuations Proc. 14th Int. Symp. on Plasma Chemistry
1997 Influence of the anode material on an argon arc IEEE (Prague, Czech Republic) ed M Hrabovsk’y et al (Prague:
Trans. Plasma Sci. 25 891–6 Institute of Plasma Physics, CAS) pp 233–8
[34] Bauchire J M, Gonzalez J J and Gleizes A 1997 Modeling of a [57] Duan Z and Heberlein J 2002 Arc instabilities in a plasma
dc plasma torch in laminar and turbulent flow Plasma spray torch J. Thermal Spray Technol. 11 44–51
Chem. Plasma Process. 17 409–32 [58] Inaba T, Oi H, Ito A and Endo M 1997 A trial calculation of
[35] Amakawa T, Jenista J, Heberlein J and Pfender E 1998 young’s modulus of Ar torch-plasma for air cross-wind
Anode-boundary-layer behaviour in a transferred, comparing with measured values Proc. 12th Int. Conf. on
high-intensity arc J. Phys. D: Appl. Phys. Gas Discharges and their Applications (Greifswald,
31 2826–34 Germany) ed G Babucke (Greifswald: Institute of Low
[36] Tanaka M and Ushio M 1999 Observations of the anode Temperature Plasma Physics) pp 115–18
boundary layer in free-burning argon arcs J. Phys. D: Appl. [59] Tanaka S I, Ito A, Kameda T, Iwao T, Inaba T and Miyashita Y
Phys. 32 906–12 2000 Voltage gradient of torch plasma forced with
[37] Tanaka M and Ushio M 1999 Plasma state in free-burning high-speed cross wind Proc. 13th Int. Conf. on Gas
argon arc and its effect on anode heat transfer J. Phys. D: Discharges and their Applications (Glasgow, UK)
Appl. Phys. 32 1153–62 ed I D Chalmers (Glasgow: University of Strathclyde)
[38] Tanaka M, Ushio M and Wu C S 1999 One-dimensional pp 605–8
analysis of the anode boundary layer in free-burning argon [60] Hartmann R M and Heberlein J V 2001 Experimental
arcs J. Phys. D: Appl. Phys. 32 605–11 investigation of anodic arc attachment instabilities Proc.
[39] Addona T, Proulx P and Munz R J 2000 Mathematical 15th Int. Symp. on Plasma Chemistry (Orleans, France)
modeling of silica anode decomposition Plasma Chem. ed A Bouchoule et al (Orleans: GREMI, CNRS, University
Plasma Process. 20 521–53 of Orleans) pp 497–502
[40] Maruzewski P, Martin A, Reggio M and Trépanier J Y 2002 [61] Iwao T, Cronin P, Bendix D and Heberlein J V R 2005 Anode
Simulation of arc–electrode interaction using sheath attachment stability and anode heat transfer for
modelling in SF6 circuit-breakers J. Phys. D: Appl. Phys. high-intensity arcs with lateral gas flow IEEE Trans.
35 891–905 Plasma Sci. 33 1123–8

29
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

[62] Cronin P 2004 Investigation in the anode boundary layer of a [86] Nagashima K T, Watanabe T, Honda T and Kanzawa A 1988
high intensity arc MS Thesis Department of Mechanical Fluid dynamic control of heat transfer to transferred arc
Engineering, University of Minnesota anode Proc. Japanese Symp. on Plasma Chemistry (Tokyo,
[63] Yang G, Cronin P, Heberlein J and Pfender E 2006 Japan) ed T Goto (Nagoya: Nagoya University) p 221–6
Experimental investigations of the anode boundary layer in [87] Menart J A 1996 Theoretical and experimental investigations
high intensity arcs with cross flow J. Phys. D: Appl. Phys. of radiative and total heat transfer in thermal plasmas PhD
39 2764–74 Thesis Department of Mechanical Engineering, University
[64] Yang G and Heberlein J 2007 The anode region of high of Minnesota
intensity arcs with cold cross flow J. Phys. D: Appl. Phys. [88] Yang G and Heberlein J 2008 Anode heat transfer by electron
40 5649–62 current and electron conduction in an atmospheric pressure
[65] Heberlein J V and Pfender E 1977 Investigation of the anode argon arc Proc. Int. Conf. on Electrical Contacts (Saint
boundary layer of an atmospheric pressure argon arc IEEE Malo, France) ed N B Jemaa (Rennes, France: Université
Trans. Plasma Sci. PS-5 171–80 de Rennes) pp 313–16
[66] Yang G and Heberlein J 2007 Anode attachment modes and [89] Morrow R and Lowke J J 1993 A one-dimensional theory for
their formation in a high intensity argon arc Plasma Sources the electrode sheaths of electric arcs J. Phys. D: Appl. Phys.
Sci. Technol. 16 529–42 26 634–42
[67] Haidar J 1997 Departures from local thermodynamic [90] Lowke J J, Morrow R and Haidar J 1997 A simplified unified
equilibrium in high-current free burning arcs in argon J. theory of arcs and their elctrodes J. Phys. D: Appl. Phys.
Phys. D: Appl. Phys. 30 2737–43 30 2033–42
[68] Leveroni-Calvi E 1990 Electric probe measurements in the [91] Lowke J J. and Quartel J C 1997 Use of transport coefficients
boundary layer of thermal arcs: theory and experiments to calculate properties of electrode sheaths of electric arcs
PhD Thesis University of Minnesota Aust. J. Phys. 50 539–52
[69] Busz-Peuckert G and Finkelnburg W 1955 Die Abhängigkeit [92] Lago F, Gonzalez J J, Freton P and Gleizes A 2004 A
des Anodenfalles von Stromstärke und Bogenlänge bei numerical modelling of an electric arc and its interaction
Hochtemperaturbögen Z. Phys. 140 540–6 with the anode: I. The two-dimensional model J. Phys. D:
[70] Busz-Peuckert G and Finkelnburg W 1956 Zum Appl. Phys. 37 883–97
Anodenmechanismus des thermischen Argonbogens Z. [93] Chen D M, Hsu K C and Pfender E 1981 Two-temperature
Phys. 144 244–51 modeling of an arc plasma reactor Plasma Chem. Plasma
[71] Schoeck P and Eckert E R G 1961 An investigation of anode Process. 1 295–314
heat transfer in high intensity arcs Proc. 5th Int. Conf. on [94] Hsu K C and Pfender E 1983 Analysis of the cathode region of
Ionization Phenomena in Gases (Munich, Germany) a free-burning high intensity argon arc J. Appl. Phys.
(Amsterdam: North-Holland) pp 1812–29 54 3818–24
[72] Schoeck P and Maisenhälder F 1966 Zur Druckabhängigkeit [95] Nemchinsky V A 1994 Plasma parameters near a small anode
der Anodenfallspannung Beiträge Plasmaphys. 5/66 in a high-pressure arc (gas metal arc welding) J. Phys. D:
345–64 Appl. Phys. 27 2515–21
[73] Hemmi R, Yokomizu Y and Matsumura T 2003 Anode-fall [96] Mentel J and Heberlein J 2010 The anode region of low
and cathode-fall voltages of air arc in atmosphere between current arcs in high intensity discharge lamps J. Phys. D:
silver electrodes J. Phys. D: Appl. Phys. Appl. Phys. 43 023002 (this issue)
36 1097–106 [97] Benilov M S 2008 Understanding and modeling
[74] Dickson D J and von Engel A 1967 Resolving the electrode plasma–electrode interaction in high pressure arc
fall spaces of electric arcs Proc. R. Soc. 300 319–25 discharges: a review J. Phys. D: Appl. Phys. 41 144001
[75] Oberth R C and Jahn R G 1972 Anode phenomena in [98] Almeida N A, Benilov M S, Hechtfischer U and Naidis G V
high-current accelerators AIAA J. 10 86 2009 Investigating near-anode plasma layers of very
[76] Leveroni E and Pfender E 1991 Investigation of the anode high-pressure arc discharges J. Phys. D: Appl. Phys.
boundary layer of free-burning high intensity arcs Proc. 42 045210
Welding and Joining Processes, ASME Winter Annual Mtg [99] Rat V, Murphy A B, Aubreton J, Elchinger M F and
(Atlanta, GA) Fauchais P 2008 J. Phys. D: Appl. Phys. 41 183001
[77] Yang G and Heberlein J 2007 Instabilities in the anode region [100] Iwao T, Cronin P, Bendix D and Heberlein J 2003 Anode
of atmospheric pressure arc plasmas Plasma Sources Sci. attachment stability and anode heat fluxes for high intensity
Technol. 16 765–72 arcs with argon and nitrogen gas flow parallel to the anode
[78] Sanders N A 1979 The effect of anode evaporation on the Proc. 16th Int. Symp. on Plasma Chemistry (Taormina,
behavior of a high intensity arc MS Thesis University of Italy) (Bari, Italy: University of Bari) ed R d’Agostino et al
Minnesota unpaginated CD
[79] Pfender E 1999 Thermal plasma technology: where do we [101] Li H-P, Heberlein J and Pfender E 2005 Three-dimensional,
stand and where are we going? Plasma Chem. Plasma non-equilibrium effects in a high-intensity blown arc IEEE
Process. 19 1–31 Trans. Plasma Sci. 33 402–3
[80] Jenista J, Heberlein J and Pfender E 1996 Model for anode [102] Baudry C, Vardelle A and Mariaux G 2005 Numerical
heat transfer from an electric arc Proc. 4th Int. Thermal modeling of a dc non-transferred plasma torch: movement
Plasma Processes Conf. (Athens, Greece) ed P Fauchais of the arc anode attachment and resulting anode erosion
(New York: Begell House, Inc.) pp 805–13 High Temp. Mater. Process. 9 1–18
[81] Nestor O H 1962 Heat intensity and current density [103] Colombo V and Ghedini E 2005 Time dependent 3-D
distributions at the anode of high current, inert gas arcs J. simulation of a non-transferred arc plasma torch: anode
Appl. Phys. 33 1638–48 attachment and downstream region effects Proc. 17th Int.
[82] Eberhart R C and Seban R A 1966 The energy balance for a Symp. on Plasma Chemistry (Toronto, Canada)
high current argon arc Int. J. Heat Mass Transfer ed J Mostaghimi et al (Toronto: University of Toronto)
9 939–40 unpaginated CD
[83] Lawton J and Mayo P J 1972 J. Phys. D: Appl. Phys. 5 79 [104] Trelles J P, Pfender E and Heberlein J 2006 Multiscale finite
[84] Bulanyi P F and Polyakoo S P 1981 High Temp. 19 351 element modeling of arc dynamics in a DC plasma torch
[85] Tsai N S and Eager T W 1985 Metall. Trans. B 16 841 Plasma Chem. Plasma Process. 26 557–75

30
J. Phys. D: Appl. Phys. 43 (2010) 023001 Topical Review

[105] Trelles J P, Heberlein J V R and Pfender E 2007 [108] Terasaki A, Tanaka M and Ushio M 2003 Effects
Non-equilibrium modeling of arc plasma torches J. Phys. of anode heat transfer on weld penetration in
D: Appl. Phys. 40 5937–52 gas tungsten arc welding Trans. JWRI 32
[106] Chazelas C, Moreau E, Mariaux G and Vardelle A 2006 29–31
Numerical modeling of arc behavior in a dc plasma torch [109] Redwitz M, Dabringhausen L, Lichtenberg S,
High Temp. Mater. Process. 10 393–406 Langenscheidt O, Heberlein J and Mentel J 2006
[107] Moreau E, Chazelas C, Mariaux G and Vardelle A 2006 Arc attachment at HID anodes: measurements
Modeling of the restrike mode operation of a dc and interpretation J. Phys. D: Appl. Phys.
plasma spray torch J. Thermal Spray Technol. 15 524–30 39 2160–79

31

You might also like