Surface Discharge Ignition of an Electrospray
Parker Gray, Alexei Saveliev
North Carolina State University
Abstract: Electrospraying occurs when a conductive fluid is charged to a high voltage and flows out of a small
nozzle. The structure of the spray depends on the conductivity and surface tension of the fluid. The spraying
characteristics of water, ethanol, and kerosene were explored. A spray of droplets forms more easily as surface
tension decreases and conductivity increases. Such a spray may be used as an atomization mechanism.
Breakdown along the surface of a fluid jet or through a fluid spray occurs at a lower voltage than does breakdown
in air [1]. This phenomenon is explored as a potential ignition source for an ethanol spray and a kerosene jet.
Ignition occurs most often at the charged nozzle several milliseconds after breakdown. The ignition is due to a
continuous discharge column through the spray or jet. The most likely cause of ignition is heating at the nozzle.
For successful ignition, the discharge has a minimum duration, minimum energy, and maximum power so as to
provide sufficient heating to ignite the fuel without dissipating too much of the supplied energy as a shock.
Successful ignition was achieved both for kerosene and for ethanol. The result is a single mechanism that is
capable of both producing a spray and acting as an ignition source.
Keywords: Electrospray, Fluid Jet Discharge, Ignition
1. Introduction
Electrosprays have been widely studied at low flow
rates, but macroscopic sprays with the same
characteristics have not [2-4]. These larger sprays,
which are driven by the pump’s power rather than
purely by electrical forces, can be applied to ignition
and combustion systems. These sprays also have the
advantage that such a spray reduces the necessary
breakdown voltage through the spray as compared to
breakdown in air, meaning that a discharge may be
generated across a larger gap width [1]. This has the
potential to allow for the development of an ignition
system in which ignition can be achieved through
the center of a spray, with the same mechanism used
for both atomization and ignition.
2. Experimental Setup
The basic setup (see Figure 1) began with the fluid,
which could be tap water, distilled water, ethanol, or
kerosene, being pumped to the stainless steel needle,
which had an inner diameter of 0.6 mm. The needle
could be charged by the high voltage supply to up to
+30 kV at 10 mA. The fluid then flowed down out
of the needle and through the grounded stainless
steel ring electrode, which had a diameter of 13.5
mm. Below this electrode was the grounded
collection pan.
Figure 1: Experimental Setup.
For flow characterization, the camera used
was an AVT Stingray. The CH filter, intermediate
resistor between the power supply and needle
electrode, the high voltage probe, and the current
monitor were not used for flow characterization.
The camera captured about 15 images at each flow
rate and voltage setting, using every other image that
the camera provided at a frame rate of 15 frames per
second. The reported voltages for these images were
the voltages displayed by the power supply.
The ignition experiments utilized all of the
pieces seen in Figure 1 and replaced the AVT
Stingray camera with an intensified Kodak Ektapro
HS 4540. The CH filter was used here to eliminate
much of the light from the discharge while keeping
that of the flame visible, effectively compressing the
brightness of everything observed into a much
narrower range. For the ignition trials, the voltage
was initially set to a voltage higher than the
minimum necessary to induce breakdown and then
turned off. Next, the fluid flow was turned on with
the voltage off. Once the fluid flow had been given
sufficient time to stabilize (10-20 seconds was
sufficient), the voltage was switched on. The trigger
point for the oscilloscope was set to be the voltage
drop due to this discharge. Thus, the actual voltage
at which breakdown occurred, which was lower than
the voltage at which the power supply had been set,
could be measured. The intermediate resistor was
needed to limit the amount of current provided by
the power supply such that the discharge had a low
enough power to ignite the mixture rather than
simply dissipating its energy as a shock. This
resistor could be varied between 100 kΩ – 50 MΩ,
with most trials performed at 1 MΩ.
dripping, capillary breakup region and the spiraling
region. It should be noted also that the transition
and spiraling regions occurred at lower voltages for
tap water than for distilled water. In addition,
distilled water formed a much clearer, longer lived
spiral than did tap water. This provides the first
evidence that conductivity plays a role in
determining the structure of such a spray, all other
things being equal.
3: Results
Figure 2: Distilled water electrospray.
voltage. Right: 10 kV applied voltage.
Flow characterization observations revealed that tap
and distilled water’s sprays had three distinct
regimes. At lower voltages, the fluid jets acted
much like they did under no voltage and simply
broke up into droplets after a particular time. The
difference under voltage was that the time that it
took to break up, and thus also the jet length,
decreased with increasing voltage, albeit rather
slowly (see Figure 1, left). Next, as voltage
increased, a transition regime was observed in which
the droplets began to spread apart in a cone shape
instead of falling straight down. The breakup time
increased noticeably faster during this regime (see
Figure 1, right). In the third regime, the spray
stabilized into a spiraling pattern. The jet was held
together by surface tension here, finally shearing
into droplets after having formed a clear spiraling
cone that gave rise to the droplet spray pattern (see
Figure 2, left). This spiraling pattern gives a clue as
to the nature of the conic spray seen in the transition
region. One can see the beginning of such a spiral in
the form of a hook that is the infant form of such a
spiral (see Figure 1, right). The transition regime is
thus simply the unstable region between the stable
Left: 6 kV applied
Ethanol’s spraying characteristics differed
from those of either water solution. At the lowest
voltages, dripping with largely stable length was
once again observed. Once electrical forces did
begin to play a more significant effect on the flow,
what was seen was similar to the transition region
seen in water. The jet began to break into a conic
shaped spray of droplets with a small hook at the
end of the jet itself. Ethanol deviated from the
behavior seen by water as voltage continued to
increase; rather than transitioning into a second
stable region characterized by a spiral shaped jet, the
ethanol spray continued to decrease in length until it
almost had reached the needle electrode (see Figure
3, right). Ethanol’s lower surface tension relative to
water seems to prevent the formation of a spiral.
Also, in spite of the fact that ethanol has lower
conductivity than distilled water, breakup began at
lower voltages for ethanol than for water. It should
be noted a corona discharge has been observed for
ethanol at 6-9 kV in this setup, albeit with reduced
current relative to when no fluid was present. No
corona was observed in that range for water [5]. The
electrical forces acting on the ethanol therefore may
be stronger than those acting on water due to the
presence of space charge from this corona discharge.
These greater forces would then facilitate breakup.
Figure 4: Kerosene electrospray. Left: 6 kV applied voltage.
Right: 23 kV applied voltage.
Figure 3: Left: Distilled water electrospray, 13 kV applied
voltage. Right: Ethanol electrospray, 15 kV applied voltage.
Kerosene, which had conductivity too low to
be measured, did not form a spray at the voltages
tested. The kerosene jet experienced no visible
effects due to the electric field at lower voltages (see
Figure 4, left), then made a quick transition from its
jet length at no voltage to a much shorter jet length
at around 15 kV. The jet still flowed directly
downward at these voltages and the only visible
electrical effect was a reduction in length. At the
highest voltages tested, one can see the path of the
jet finally begin to change due to the presence of the
electric field. A hook may be seen at the bottom of
the jet that breaks up into droplets. These droplets
appear to spiral about a central axis, albeit much less
pronouncedly than was seen for the other fluids (see
Figure 4, right). Kerosene carried no appreciable
surface charge at voltages below 10 kV. In fact, no
current was detected for a kerosene flow at these
voltages that could not be explained by a slight
suppression of the corona discharge [5]. Some
conductivity, then, is necessary for appreciable
electrical effects to be seen on such a fluid jet and
especially to produce a spray if voltages are to be
kept below those at which more sparking begins to
occur.
Ignition was achieved both for ethanol and
for kerosene. The voltage characteristics and images
from the video taken for one such ignition may be
seen in Figures 5 & 6. This video was taken for an
ethanol spray at a jet velocity of 0.91 m/s, the same
flow rate as for all the images seen in Figures 2-4.
Ignition occurred just after the end of these voltage
measurements, at about 4-5 ms after the initial spark
and can be seen as the thicker fuzzy part of the
discharge column in frames (e) and (f) of Figure 6.
The initial spark’s energy is estimated to be 6 mJ,
the energy held by line capacitance. It is very
unlikely that this is sufficient energy for ignition [6],
so the ignition must be due to the continuous
discharge column that follows this initial spark.
This discharge column, which is a surface discharge
hopping between droplets similar to those seen by
Shmelev [1], can be seen as the luminous column in
Figure 6. This column is supported by a voltage of 5
kV, insufficient to initiate any more than the weakest
coronas seen but enough to support the flow of
charge once the breakdown channel has been
created. The energy deposited by this discharge
column before ignition may be estimated from the
current that passed through the 1 MΩ resistor
between the 30 kV supply and the 5 kV needle, and
is equal to 500 mJ. Once ignition does occur, the
flame can be seen to begin at the needle electrode
and fall down along the jet. Ignition is most likely
due to heating at the needle tip due to the continuous
discharge column. It was observed that the flame
must begin at one of the electrodes in order to
survive because the fuel flow speed was greater than
the laminar flame speed. Any flame that began in
the gap simply fell out of the gap and was snuffed in
the collection pan.
spark across the kerosene spray, due to its lack of
conductivity, was only achievable at even shorter
gap widths (19 mm for kerosene as opposed to 15
mm for ethanol). The supply thus had to be set
lower to prevent sparking when no fluid was present.
4: Conclusions
Figure 5: Voltage characteristics for ethanol ignition event seen
in Figure 6.
Macroscopic charged sprays mimic the
behavior of those seen in the microscopic sprays
studied previously [2-4]. A jet under electrical
forces exists at low voltages as a jet breaking up into
droplets, similar to the no voltage case. If surface
tension is sufficiently high, the spray will transition
into a spiraling jet. Electrical effects, i.e. jet breakup
and spray production, increase with increasing
conductivity. Some conductivity is necessary for the
production of a spray. Droplet production seems
better with decreasing surface tension due to the
reduction in the spiraling of the jet.
Ignition by a surface discharge through this
type of a spray was achieved both for an ethanol
spray and for a kerosene jet. The source of ignition
was a continuous discharge column along the
channel formed by the initial spark along the surface
of the fluid. This discharge column heated the fuel
at the needle tip and resulted in thermal ignition after
a few microseconds.
(a) 220 us
(b) 1320 us
(c) 2420 us
References
[1] V. Shmelev, IEEE Transactions on Plasma
Sciences 36, 2008, pp. 2252-2527.
[2] G. Taylor. Proc. Roy. Soc. London 313(1515),
1969, pp. 453-475.
(d) 3520 us
(e) 4620 us
(f) 5720 us
Figure 6: Ignition of ethanol. Times are time elapsed after
initial spark.
Kerosene ignition was achieved using a
continuous discharge column. Ignition was achieved
for kerosene in half the time necessary for ethanol,
with a flame clearly visible after 2.2 ms. The energy
deposited before ignition was thus reduced to 50 mJ,
in part due to decreased time before ignition and in
part due the fact that the supply was set lower. A
[3] A. L. Yarin, S. Koombhongse, D. H. Reneker,
Journal of Applied Physics 89(5), 2001, pp. 30183026.
[4] J. F. de la Mora, I. G. Loscertales, Journal of
Fluid Mechanics 260, 1995, pp. 155-184.
[5] P. Gray, Ignition of a Liquid Fuel Jet, 2010,
http://repository.lib.ncsu.edu/ir/handle/1840.16/6644
[6] K. V. L. Rao, A. h. Lefebvre, Combustion and
Flame 27, 1976, pp. 1-20.