Downloaded from http://rsos.royalsocietypublishing.

org/ on October 12, 2017

The Electric Honeycomb; an

rsos.royalsocietypublishing.org
investigation of the Rose
window instability
Muhammad Shaheer Niazi
Research
Lahore College of Arts and Sciences, Lahore, Punjab, Pakistan
Cite this article: Niazi MS. 2017 The Electric
MSN, 0000-0002-2782-3555
Honeycomb; an investigation of the Rose
window instability. R. Soc. open sci. 4: 170503.
http://dx.doi.org/10.1098/rsos.170503 The Rose window instability is a little-explored electro-
hydrodynamic instability that manifests when a layer of low-
conducting oil is placed in an electric field generated by
corona discharge in a point-to-plane configuration. Above a
Received: 24 May 2017 critical voltage, the instability starts as a single dimple in the
oil layer right below the point electrode and subsequently
Accepted: 4 September 2017
evolves into a characteristic pattern of polygonal cells. In this
study, we experimentally explore governing parameters that
guide the instability and document geometric attributes of
the characteristic cellular pattern. The driving force for the
Subject Category: instability has been attributed to the buildup of charged ions
Physics which in turn apply an electric pressure on the oil surface. We
confirm the charged surface distribution using thermal imaging
and demonstrate that the instability can be locally inhibited by
Subject Areas:
preventing charge buildup under an ion shadow.
fluid mechanics

Keywords:
instability, electrohydrodynamics, 1. Introduction
Rose window
Controlled spreading of droplets and liquid films over a solid
substrate has widespread use in applications such as coating,
printing, micro- and bio-fluidics to name a few. Recently, it has
Author for correspondence: been proposed that corona discharge can be an effective technique
for controlled droplet spreading [1,2]. Corona discharge is effected
Muhammad Shaheer Niazi
by the ionization of air in the vicinity of a sharp-tipped conductor
e-mail: niazi.muhammadshaheer
carrying a high voltage. Free charges created in the process
[email protected] accumulate atop the droplet that rests on a conducting plate
and subsequently help guide the motion of the droplet under
an applied electric field. However, as the droplet spreads to a
film, the air–liquid interface can be rendered unstable and under
certain conditions spontaneously evolves into a stable pattern
of polygonal cells, each of a size several times larger than the
thickness of the film (figure 1). The instability was nicknamed the
Rose window instability (RWI) by Pérez [3], who first identified
that it occurs only in low-conducting liquids. Subsequently, the
RWI has been explored in more detail both experimentally and
theoretically by only a handful of researchers. Vega & Pérez
[4] used linear stability analysis to obtain a theoretical estimate

2017 The Authors. Published by the Royal Society under the terms of the Creative Commons
Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted
use, provided the original author and source are credited.
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
Figure 1. Close up of the Rose-window instability (RWI) as observed in our experiments.

of the critical voltage at which RWI initiates for an air–ohmic liquid interface subject to a perpendicular
electric field. The results were verified against experiments in a subsequent study [5], wherein the
RWI was investigated experimentally for low-conducting fluids in both the ohmic and non-ohmic
regimes. More recent theoretical investigations have extended the stability analysis to consider the role
of liquid motion [6] and the presence of an insulating liquid [7]. In this study, we extend experimental
investigation of the RWI and explore the parametric space wherein the instability manifests itself. We
also document geometric attributes of the instability pattern and finally prove the role of charged ions as
the key ingredient that drives the instability.

2. Material and methods

Early research into the RWI [3] identified the key conditions needed for the instability to reveal itself: a
fluid with very low conductivity which makes it a dielectric barrier to the charge that aspires to flow to
the underlying conducting plate; a point electrode to have a high curvature, like a needle point, to create
an injection electrode that can ionize air molecules; and a pulsed DC high voltage supply (EMCO Q101-
5). The experimental set-up for this study consists of a custom-made wood table to support the plate
electrode (figure 2a). The injection electrode is a sandpapered steel needle insulated on the body to keep
away stray fields and corona leakage (figure 2b). The needle is hoisted on a burette stand and a vernier
scale used to measure adjusted pin heights. The plate electrode consists of an FTO (fluorinated tin oxide)
coated glass electrode [4] which aids in imaging. A toroidal wooden collar pasted on the glass electrode
helps confine the low-conducting fluid, which for our study is machine oil (approx. 890 kg m−3 ). The
high voltage supply used for the experiments provides variable voltage in the range of 0–10 kV DC at
75–500 kHz. Voltage and current were measured using UNI-T UT60 and an EMCO V1G HV divider
(1000 : 1). Schlieren photography using shadowgraphy was also done using a 5.1 inch parabolic mirror.

3. Results
3.1. Critical voltage
We investigate the critical voltage at which the RWI manifests itself as a function of the pin height and
liquid layer thickness. Here, we define the start of the instability as the formation of a single dimple in the
oil layer (figure 3) which is readily detected by placing a piece of paper under the glass electrode. Owing
to the refractive properties of the dimple, a shadow is formed on the paper signifying its presence.
To investigate the role of pin height, we start with an oil layer of thickness approximately 1 mm
which is subsequently held constant. Next, we increment the pin height from 1 to 5 cm (using a step
of 1 cm) and measure the voltage at which the instability initiates. We find that the critical voltage
increases linearly with an increase in pin height (figure 4). This trend is in accord with experimental
results reported by Pérez [3], who attributed the instability to the electric pressure acting at the air–liquid
interface. With increasing pin height L, the voltage V needed to obtain the same strength of electric field
E (and correspondingly electric pressure P ∼ E2) must increase linearly in accord with E ∼ V/L as we can
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

(a) (b) 3

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
Figure 2. Experimental set-up. (a) A glass plate coated with FTO serves as the plate electrode while a cylindrical barrier prevents oil
leakage. (b) Needle body insulated with heat sink tubing, to minimize stray fields and corona leakage.

Figure 3. A surface dimple which marks the initiation of the RWI. Wooden collar is also visible.

assume the electric field at the oil surface to be homogeneous due to a small ratio of the pin height and
instability radius. Using an I–V graph like in figure 5, one can also determine critical voltage versus pin
height.
Next, we investigate the role of the liquid film thickness d (figure 6). Previous studies have explored
the variation of critical voltage with d within a narrow range of values from d ∼ 0.6 to 1.6 mm. Here, we
explore a much larger thickness range from d ∼ 1.9 to 6.8 mm. The thickness was gradually increased by
adding a certain amount of oil and measuring the depth using a vernier caliper. The pin height was kept
constant at 2.8 cm. We find that even for the extended range of values the critical voltage V increases
roughly linearly with d in keeping with incomplete results in fig. 7 of Pérez [3], where the critical voltage
increases from approximately 2 mm to approximately 2.5 mm with ‘point positive’. So we note that after
a thickness of approximately 2 mm, the critical voltage starts increasing as shown by our results and two
sets of values in Pérez [3].

3.2. Geometric characteristics of the Rose window instability

We explore geometric characteristics of the RWI for our point–plane configuration which, to our
knowledge, has not yet been explored in detail. We observe that if the voltage is held constant, beyond
the critical voltage, the unstable air–liquid interface settles into a stable configuration of polygonal cells
with a near circular boundary. We explore the spatial extent of the cell pattern, measured as the radial
distance from the pin to the edge of the cell pattern, which we refer to as the instability radius (figures 7
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

10
4
9

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
8
7

voltage (kV) 6
5
4
3
2
1
0
0 1 2 3 4 5 6
pin height (cm)

Figure 4. Variation of critical voltage with pin height. The dashed line shows a linear-fit to the experimental data. Horizontal error bars
in accordance to error factor of the vernier caliper. Vertical error bars too small to be seen.

I–V graph
10

4
current (µA)

0
–15 –10 –5 0 5 10 15
–2

–4

–6

–8

–10
voltage (kV)
experimental experimental (–)

Figure 5. Current–voltage characteristic for the RWI instability. Note that no current is recorded before the threshold voltage is reached
which marks the initiation of the instability/corona. Pin height 1 cm, layer thickness roughly 2 mm.

critical voltage (experimental) with reference to layer thickness/pin at 2.8 cm

9
8
7
voltage (kV)

6
5
4
3
2
1
0
1 2 3 4 5 6 7 8
layer thickness (mm)
critical voltage (experimental) linear (critical voltage (experimental))

Figure 6. Variation of critical voltage with liquid layer thickness. The dashed line shows a linear-fit to the experimental data. Horizontal
error bars in accordance to error factor of the vernier caliper. Vertical error bars too small to be seen. Pin height 2.8 cm.
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

3.5
5
3.0

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
instability radius (cm)

2.5

2.0

1.5

1.0

0.5

0 2 4 6 8 10
voltage (kV)

Figure 7. Instability radius versus injection voltage. The triangles represent data for a pin height of 1 cm and the diamonds represent
data for a pin height of 2 cm. Dashed lines are best-fit curves.

(a) (b) (c)

Figure 8. Evolution of the RWI with increasing voltage beyond the critical threshold. These images were obtained with a pin height of
1 cm and the corresponding threshold voltage is 2.95 kV. The paper placed below the system is a means to view the instability better. The
lines are merely marks to record size. (a) RWI at around 3.2 kV. (b) With increasing voltage (approx. 5 kV), the instability develops with a
defined, yet irregular, boundary. (c) At 10 kV, the cells reduce to tiny polygons within a more well defined circular instability radius.

and 8). We measure the instability radius as a function of the applied voltage for two different pin heights.
For each case, the instability radius increases rapidly at first, but soon saturates to a near constant value
(figure 7). We attribute this response to the fact that the electric field lines emanating from the injecting
electrode spread at a certain deviation angle from each other, which in turn leads to an inhomogeneity of
the electric field that acts on the surface of the oil. As we increase the voltage, the field density increases
which acts to reduce the inhomogeneity of the electric field as neighbouring lines deviate less from each
other.
Radial heterogeneity in the electric field strength also manifests in the size of the polygonal cells that
form due to RWI. In the tip–plane configuration, the electric field is strongest right underneath the tip
and reduces radially away from the tip. Correspondingly, we find that the cell sizes vary radially with
the smallest cells below the tip and larger cells at the boundary (figure 9). Cell sizes would be related to
field density owing to the gradient of electrostatic pressure on the oil due to the ionic bombardment.

3.3. Surface charge distribution

The driving force for the RWI has been attributed to the action of the electric field on the surface charges
that accumulate atop the low-conducting oil layer. As the charged ions bombard the surface, they carry a
certain energy, part of which must be deposited onto the surface of the oil and over time should lead to a
significant shift in the oil temperature. Thus, to reveal the distribution of these surface charges, we resort
to thermal imaging. Using a thermal camera, we take images of the oil surface at four instances: one at the
start, then at half a minute and a minute after the initiation of the instability and another after about 5 min
(figure 10). The radial temperature gradient evident in these images serves to verify the assumption of a
radially inhomogeneous charge distribution, which in turn manifests radial variations in the geometric
attributes of the polygonal cells as discussed earlier. To further strengthen the role of surface charges as
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

0.6
6

0.5

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
0.4
bead size (cm)

0.3

0.2

0.1

0
centre boundary

Figure 9. Radial variation of cell size. The radial distance is measured from the tip of the injecting electrode. Pin height 3 cm at 10 kV.

the key ingredient responsible for the RWI, we shield part of the oil surface from charge accumulation
by creating an ion shadow. An ion shadow forms when an object blocks the path of the charged ions
towards the surface of the oil. Even thin objects, like paper, are sufficient to produce a sharp shadow
that inhibits charge accumulation on the oil surface. In the absence of surface charges the instability is
locally inhibited, as evident from figure 11. Even metal objects produce the same effect when placed in
the path of the charged ions. Using Schlieren photography (figure 12), we imaged the corona streamer
which shows that the streamer and gas flow stay roughly the same size as the needle diameter, while the
instability continues to occur at a larger radius.

3.4. Current–voltage characteristic

Lastly, we investigate the evolution of the I–V curve in our experimental set-up. We note that the
polygonal cells of the RWI form to allow for the transfer of the charged ions atop the oil to the
grounded plate electrode. Thus, one might visualize the instability as a breakdown of the dielectric
barrier of oil which prevents electrical conduction between the injecting electrode and the grounded
plate. Our measured I–V curve confirms this notion. We find that before the instability initiates, we
measure zero current at the grounded electrode. As the instability forms, ions are able to migrate to the
grounded electrode and we correspondingly measure an increase in current. With increasing voltage,
the depressions created by the RWI grow rapidly, allowing more ions to ground and correspondingly we
record an exponential increase of the I–V curve.

4. Discussion
If we compare the RWI to other instabilities, we see that they have something in common, that is to
achieve an equilibrium in the system. The RWI achieves a flow of current out of the system by essentially
completing the circuit. It does that by creating the polygonal patterns. If we look at the case in figure 11
and instead place a wire connected to an ammeter in the flow of ions, then we notice an inflow of current
into the wire. This shows that in the system the charged ions would seek the easiest possible way to get
grounded and when no other option is available the RWI starts to form.
In figure 6, we notice the increasing voltage with respect to increasing thickness which tells us that a
higher ion energy is needed to start the instability, because the ions are farther from the plate electrode
which results in a weaker force. So higher voltages, causing higher energy ions, are needed to create a
significant force to start the instability. We also see in the thermal gradient that the higher energy ions
bombard the central region of the instability and over there we also see a smaller cell size. In figure 12, we
can view the corona streamer and the resulting gas flow. Using the schlieren method, we can also observe
an increased temperature of gas flowing towards our oil due to residual gas heating and exothermic
chemical reactions taking place due to ionization. This can explain the faster rate of temperature increase
at the centre of the instability as seen between figure 10b and c, where the central temperature increases
by 0.5°C while border temperature remains rather unchanged.
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

(a) (b)
7

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
time

DTedgeÆcentre = 1°C

(c) (d)

time

DTedgeÆcentre = 1.5°C DTedgeÆcentre = 4.8°C

Figure 10. Thermal imaging of the RWI. (a) Start of instability, almost uniform temperature gradient (18.6°C). (b) Half a minute after
initiation. (c) Image obtained about a minute after initiation of the instability. (d) Image obtained about 5 min after initiation of the
instability. For each case, point 2 marks the centre of the instability which occurs right underneath the tip of the injecting electrode. For
both (c) and (d), a radial temperature gradient is evident with temperatures rising as we traverse from the centre to the boundary of the
instability, roughly marked by points 2 and 3.

Figure 11. Local inhibition of the RWI by creation of an ion shadow that prevents surface charge built-up around the insulating pen. The
curvature of the shadow results from the shape of the fountain-shaped electric field in the tip–plane configuration.

5. Conclusion
Instability of the air–liquid interface under an applied electric field has shown that the ions in effect of
the electric field go towards the plane electrode. Unlike an air capacitor, in our point–plane configuration
ions are produced which are charged. These ions flow towards the plate to be grounded but the dielectric
layer in effect prevents them from doing so. Before the instability can occur, the charges accumulate on
 Downloaded from http://rsos.royalsocietypublishing.org/ on October 12, 2017

rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170503

................................................
Figure 12. Schlieren images of the corona streamer emanating from the point electrode (needle). Using schlieren photography we can
visualize variations in density/temperature in fluid mediums such as air in our case.

the surface of the oil and in effect of their relative energies, produce a certain force on the oil which when
strong enough pushes the oil out of the way, This initial opening or weak spot in the oil is our initial
dimple and this opening weakens further, increasing in size. The process of self-organization occurs in
the oil to make the openings in essence the boundaries of polygons, which ensures stability of the system.
As we see in figure 5, just as the dimple forms at the critical voltage, a current is detected in our system
and that current increases as we increase voltage. If we recall from figures 7 and 8 we know that the
cell size gets smaller and smaller with increased voltage which means more surface area for the ions to
ground, and that supports our conclusion of the mechanism of the RWI. Furthermore, we also conclude
that the ions are the main cause of the instability and image the system in two new perspectives (thermal
and schlieren) which reveal previously unknown phenomena associated with the RWI.
Data accessibility. Data are available at Dryad (http://dx.doi.org/10.5061/dryad.q680n) [8].
Competing interests. The author declares no competing interests.
Funding. Funding was provided by COMSATS Institute of Information Technology, Islamabad, Pakistan and Lahore
University of Management Sciences (LUMS), Lahore, Pakistan for preparation of the International Young Physicists
tournament 2016 by team Pakistan.
Acknowledgements. I thank Dr. Farida from COMSATS and Dr. Sabeih from LUMS for supporting the research and
providing access to laboratories to conduct research. I thank all the staff in the LUMS Physics Department for
machining and setting up the equipment and doing rounds of the markets to ensure best results. I am grateful to
Dr. Troy Shinbrot and his colleague Tapan Sabuwala for helping in formatting the paper in a way that is acceptable to
the scientific community. Lastly I thank my mother, for giving me courage and patience to conduct the research and
the drive to write this manuscript.

References
1. Mahmoudi SR, Adamiak K, Castle GSP. 2013 On the 4. Vega F, Pérez AT. 2002 Instability in a to charge injection. Phys. Fluids 18, 104108.
corona discharge spreading of dielectric liquid films. non-ohmic/ohmic fluid interface under a (doi:10.1063/1.2363219)
J. Electrostat. 71, 496–498. (doi:10.1016/j.elstat. perpendicular electric field and unipolar injection. 7. Chicón R, Pérez AT. 2014 The stability of a
2012.12.038) Phys. Fluids 14, 2738–2751. (doi:10.1063/ horizontal interface between air and an
2. Mahmoudi SR, Adamiak K, Castle GSP. 2011 1.1488146) insulating liquid subjected to charge injection.
Spreading of a dielectric droplet through an 5. Vega F, Pérez AT. 2003 Corona-induced Phys. Fluids 26, 034103. (doi:10.1063/
interfacial electric pressure. Proc. R. Soc. A 467, electrohydrodynamic instabilities in low conducting 1.4868365)
3257–3271. (doi:10.1098/rspa.2011.0220) liquids. Exp. Fluids 34, 726–735. (doi:10.1007/s00 8. Niazi MS. 2017 Data from: The Electric Honeycomb;
3. Pérez AT. 1997 Rose-window instability in low 348-003-0616-y) an investigation of the Rose window instability.
conducting liquids. J. Electrostat. 40, 141–146. 6. Chicón R, Pérez AT. 2006 Instability of an interface Dryad Digital Repository. (doi:10.5061/dryad.
(doi:10.1016/S0304-3886(97)00028-4) between air and a low conducting liquid subjected q680n)