Lab-based Martian analogue experiments investigating electric and magnetic fields of dust devils

Dust devils – rotating vortices of lofted particulates formed by convection from solar heating - are a prevalent aspect of Martian weather.Within a dust devil, particles become charged though triboelectrification, and observations of terrestrial dust devils have revealed electric and magnetic fields. Understanding the behaviour of these aeolian events is vital for mission planning and duration, for example, with the Insight lander as a prime example of a mission cut short due to obscuration of solar panels by lofted dust, which is highly likely to be charged. Charged dust devils on Mars strongly imply the existence of electric discharges due to the low breakdown field. The existence of discharges would in turn imply a global electric circuit on the planet – with implications for organic chemical formation and methane losses. This work presents a laboratory-based experimental set up for investigation into both electric and magnetic fields and the feasibility of using Martian analogue dust as alternative to in-situ measurements. In a dust devil, there are two main driving motions – a vertical separation of small and large particles, and the characteristic tight spiralling motion. The experimental set-up splits these components, allowing investigation into each. It is made up of a carefully designed cylindrical tank into which dusts or powders can be dropped to allow triboelectric charging as they fall under gravity. The base of the tank has a Faraday cup for charge measurement and there is also a field mill looking into the tank from the top. The rotational component is achieved using a paddle arrangement for samples placed in the base of the tank. Electric and magnetic fields have successfully been detected from charged dust with this system.


Introduction
Mars is the fourth planet from the Sun in our Solar System, and the furthest terrestrial planet from the star.It has attracted much scientific interest due to its potential for habitability, and its proximity and similarity to Earth.Mars possesses a thin, predominately carbon dioxide atmosphere, but its defining feature is the reddish, dusty surface, shown in Figure 1.
This dust, known as regolith, is comprised of fine particulate, making it susceptible to motion induced by aeolian processes (those arising from the wind): saltation, suspension, creep and roll [1].Broadly, there are three types of dust storm on Mars -global and regional dust storms, local dust storms and dust devils.Global dust storms are large scale events, lasting for weeks or months, formed when several smaller dust storms merge.On Mars, they typically occur during the southern hemisphere summer, with a global dust storm occurring every 3-4 Mars years [2].Local dust storms occur in a specific area, and arise due to local atmospheric heating due to suspended dust, which intensifies circulation.Dust devils are smaller again, and differ to both local and global dust storms in that they have an ordered sense of rotation.For the formation of all of these storms, suspension plays a key role.
Formation of dust devils is depicted in Figure 2. The dusty surface is heated by the sun, causing adiabatic convection.The rising bubbles of air join together, forming a thermal plume, and rotation is introduced by random atmospheric eddies.This forms a thermal vortex -and surface dust is suspended by the rising plume, becoming entrained in the rotating flow -forming a dust devil.With entrained dust moving, collisions between particles are guaranteed.
In these interactions, charge is exchanged -this is a process known as triboelectrification.There are two mechanisms by which this occurs -related to the material properties and the size of the particles interacting.If the two materials are different, then charge is exchanged according to the triboelectric series -with materials higher up in the list losing electrons to materials lower down in the list [3].Metallic systems are the most well understood -charge transfer between two surface is proportional to the difference in workfunction system [4].
Between particles of the same material, charge exchange is based upon the relative size of the particles -large particles lose electrons to smaller particles, and thus small particles become negatively charged and larger particles become positively charged.The mechanism of this is less well understood than the mechanism between differing materials -it was theorised that this occurs due to all particles have the same density of surface trapped high energy electrons, and collision allows for these trapped electrons to escape to a low energy state on the other particle [4].In a dust devil, particles become stratified by mass -and assuming all particles have similar densities, this translates to sorting by size.Small, lighter particles are raised higher than larger, heavier particles, and so there becomes a separation ( d) of two oppositely charged regions, with charge q.This leads to an electric dipole ( p), Equation (1), and thus an electric field ( E), Equation (2) at a position vector R, Figure 3a.In Equation (2), ε 0 is the vacuum permittivity.
There have been several measurements of electric fields of terrestrial dust storms.In 2016, Yair et al. [5] reported electric field measurements between 1 and 8 kVm -1 of a local dust storm over the Negev in Israel.Kamra recorded dust storms with negative kVm -1 in Northern India [6] and, in the South-Eastern US, storms with -1.5 kVm -1 and +500 Vm -1 [7].The composition of the dust storm was explained as the reason for the opposing polarity of electric fields, with clay mineral dusts responsible for the negative field and silica mineral dust causing the positive fields.Another example occurred during the MATADOR desert tests in the US [8] for a dust devil that passed directly over the sensing instrumentation.This event saturated the field mill at -4.3 kVm -1 , whilst also providing, for the first time, simultaneous measurements of the magnetic field -which, in the ultra low frequency band <40 Hz, showed an uncalibrated peak [8].
Whilst the process of the formation of electric fields in charged dust is well understood, the process by which magnetic fields arise is much less well known.The example of the MATADOR campaign mentioned above is the only recorded data [8]   The suggested explanation for the magnetic field [9] was that the spiralling motion of a charged grain was analogous to a current (I) flowing in a solenoid of cross-sectional area A, with normal direction n.For a solenoid, the magnetic dipole ( µ) is given by Equation (3) and the field ( B) at a position vector R associated with this can be determined using Equation (4).The vacuum permeability is prepresented by µ 0 , Figure 3b.

Experimental Apparatus
Previous experimental work focused on electrical breakdown caused by tribocharging of dust -from agitation of sand in a low pressure carbon dioxide flask by Eden and Vonnegut [10], to Krauss et al. [11] optically detecting discharges while horizontally mixing regolith simulant (JSC-Mars-1), and dropping a mixture of regolith simulant and glass microballoons.The latter of these experiments was reproduced by Aplin et al. [12], who recommended future work try to avoid wall effects.Following previous investigation, and understanding of the process by which the electric and magnetic fields form in dust devils, an experimental apparatus was designed.The design enabled the two dominant components of motion in a dust devil to be isolated, similarly to [11], specifically, a vertical separation of charges, and a horizontal spiralling motion.The vertical separation of charges will be achieved by dropping particles from the top to the bottom of the tank [13], with expectation of generation of an electric field, but no magnetic field generated.This also offers a useful calibration and validation method, as based upon previous work [12] falling particulate has been shown to charge.The horizontal rotation will be induced using a paddle driven by hand, with previous motor driven attempts showing a large magnetic signal [14].Schematic representations of the 2 configurations can be seen in Figure 4.
One multipurpose Perspex tank is used for each mode, with internal height of 440 mm, internal diameter 340 mm and wall thickness of 13 mm.It is lined internally with grounded aluminium foil, and externally with MuMetal.MuMetal is a ferrous alloy, with high permeability, that provides a low reluctance path for magnetic flux to shield the inside of the tank from external magnetic fields.The tank is supported by an MDF base, with 4 self-levelling feet.At the top of the tank, is the dust inlet -a grounded metallic funnel -and the CS110 electric field mill.This sensor has resolution 0.32 Vm -1 , and looks down into the tank from approximately 600mm from the base, though a square hole cut into the lid.This offset distance is sufficient, to mitigate dust particles affecting the sensor, although this has also been previously used by colleagues in the Negev desert with minimal effect [5], and is readily cleanable should the need arise.It was noticed that the field mill was sensitive to external fields, so a Faraday shield, consisting of layers of aluminium foil, was installed around the aperture.A magnetometer is located in the centre of the tank.Some initial testing was performed using the engineering model of the LEMI 133 search coil from the planned ExoMars 2022 lander platform but following the revision of the mission, and the removal of the Kazachok landing platform from the planned ExoMars 2028 mission, this sensor was changed to a fluxgate magnetometer (Stefan Mayer FLC3-70).
The equipment at the base of the tank is changed between modes.In the vertical drop mode, the base consists of a Faraday cup on teflon standoffs, connected to a Keithley 6514 electrometer (Figure 4a).The voltage measured at the cup can be converted to charge, when the capacitance of the system is known.In the horizontal mixing mode, the Faraday cup is instead replaced with a non-ferrous paddle, to generate rotational motion.Separate to this work, some preliminary studies on using air to provide this driving force have proved to be successful.

Capacitance
The capacitance of the tank was determined by driving the, floating for the purpose of this measurement, internal surface of the tank with a waveform generator, outputting a triangular waveform of defined frequency and amplitude, and measuring the current (I) at the Faraday cup using the same coaxial cable set up as the other measurements, with the triaxial input of the Keithley converted to coaxial mode, the capacitance (C) can be determined, Equation (5).
The internal surface of the tank was driven with waves of differing frequencies and amplitudes.An example waveform is shown in Figure 5, with a summary of the results of each of these seen in Table 1 -where each different waveform has 2 results: one for the positive rate rising limb and one for the negative rate falling limb.The average value of the capacitance is taken as the value for the tank -28.10 pF -with a standard error of 0.15 pF.This represents a percentage uncertainty of 5%, which is carried forward to all charge measurements.

Preliminary Testing
To verify the set-up of tank, initial tests were carried out with grains that were expected to charge more than regolith simulant.

Vertical Drop
Firstly, 0.11 g of polystyrene beads (approximately 180 beads)were dropped.The Faraday cup recorded a total peak charge of magnitude 6 pC, but no signal was recorded by the field mill.It was assumed that the field was too small to detect.This low charge is likely to result from a fairly monomodal size distribution meaning little charge is exchanged in collision -in fact, it is likely that the charging that occurred is as a result of tribocharging from the walls of the funnel due to the dissimilar material [12].Next, a mixture of glass beads was used: 10 g of 2.0 mm diameter beads, and 10 g of 1.0 to 1.3 mm diameter beads.This give a bimodal size distribution of particulate, so tribocharging will occur and a bimodal sample with a broad span will charge more than a narrow span [13].
An initial drop was performed and this seemed not to show any electric field signal above the background of the instrument.It was speculated that was again due to the magnitude of the charge being too low for remote detection: a peak charge of 30 pC was detected using the Faraday cup. Figure 6 shows these results and suggests that the positively charged larger particles reach the bottom of the tank first, followed by the smaller particles, which are negatively charged.Just after the main negative peak, it can be seen that there is a small bobble, where a less negative charge is detected.This is consistent with the large particles bouncing off the the surface of the Faraday cup and recolliding with the cup again, which was observed visually.Following this, the charge decays, this is not fully shown in Figure 6.By fitting an exponential curve to this decay, the time constant can be determined to be 310.5 s, and this gives an electrical conductivity of 2.85×10 -14 Sm -1 .This is consistent with the expected conductivity of air [15].
To combat this, and provide evidence of the apparatus being successfully able to detect an electric field from the drop, the particles were pre-charged to further increase the signal.This involved putting particles inside a PVC sample pot, and then this was shaken and rubbed against a polyester/cellulose cloth.When this pre-charged mixture was dropped though the tank, an electric field was detected.The traces from both the field mill and the Faraday cup are shown in Figure 7.In one drop, the electric field peaked at 3 Vm -1 while the Faraday cup trace shows nearly 1 nC -although another test with a higher charge shows an electric field of -1 Vm -1 .This difference can be explained by the 5 Hz sampling rate of the electric field mill, and so will likely only capture the field of the dropping beads once in a single drop.The timing of this capture will not be the same across tests, and thus the charged particles configuration is likely to be different on each run.The charge profile is very different to Figure 6, showing the initial charging to be strongly positive.While there are similarities with the test for which the glass beads aren't pre-charged, the peaks have much greater magnitudes.This can likely be explained by the pre-charging process -whereby the particulate is charged based upon the difference in size, and the difference in position on the triboseries between glass and PVC.
During the drop, the magnetic field was also recorded.A substantial mains hum (magnetic field) was observed ( 250 nT at 50 Hz) inside the tank, even with the MuMetal screening.A second sensor was used outside the tank to identify and subtract it.As expected, no magnetic signal was observed for the vertical drop of the pre-charged glass beads.

Horizontal Mixing
Testing with polystyrene balls showed the rotating arm was able to provide airspeed to move the particles in a spiral inside the tank.However, when using a mixture of glass beads (as in the vertical drop Section 3.1), the rotating arm was no longer sufficient to lift the particles.Thus, polystyrene balls were used for the preliminary work in the horizontal mixing.
A hand-driven brass shaft was used to generate rotational motion while minimising the generation of magnetic signals.A portion of the spectrogram of this test is shown in Figure 8 for 0 to 200 Hz.The DC component of the magnetic field, corresponding to the Earth's magnetic field was removed through subtraction of the average of the signal.
A weak signal at frequencies <50 Hz can be seen in addition to the strong background mains hum.It is hard to pinpoint whether this is an effect of the rotation of the charged polystyrene The electric field mill was also recording in these tests, though the trace isn't presented here.In this test, some of the polystyrene balls loft higher than others, and with differing charge, an electric field can be expected.There was speculation that a field was detected at around the 30 s mark of the test -peaking twice in the region of -0.6 Vm -1 , with an otherwise constant (for the duration of the test) ± 0.3 Vm -1 .

Electric Fields
Following the success of the vertical drop test with glass beads, the medium was changed to Martian regolith simulant.This is the Exolith Lab Mars Global Simulant 1 (MGS-1) -a silicarich dust, developed as the first mineralogically accurate simulant.It is based off the soil analysis of the NASA Curiosity rover in the Gale Crater.Unlike previous simulants, it is manufactured by mixing pure minerals together in a realistic particle size distribution, rather than being mined [16].
An initial drop showed a charge on the Faraday cup, and an electric field.The trace of the drop is shown in Figure 9, with 20 g of MGS-1 it shows a peak charge of around -370 pC.Delayed from the drop by about 20 s, the electric field peaks at 2 Vm -1 .This delay can be explained by the simulant having a large fine particulate fraction, that is highly charged.By assuming a constant density (in this case, the bulk density of the material) and spherical particles, the datasheet can be used to estimate an approximate number percentage for each of the diameter bins.Table 2 shows that the sample (by number) is nearly 75% small particulate -which is known from principles of tribocharging to charge negatively.Further work is needed to better understand this delay.It should also be noted that no magnetic field over the expected background was observed.
In addition to the MGS-1 simulant, testing was also performed on the MGS-1C variant.This is a hydrated clay varient, based upon the reference case "C" from NASA's Mars Water In-Situ Resource Utilization Planning Study [18].With ExoMars' planned landing site at Oxia Planum, a large exposure of clay-bearing rocks, it is believed that this simulant would be more representative.
In the drop with 20 g of the MGS-1C simulant, it was seen that a much lower net charge than the other simulant was observed -with the Faraday cup showing only -45 pC.It can also be seen that the same delay between the drop and the electric field occurs.Between the two drops, the sign of the electric field changes -from the work of Kamra [6][7], it was expected that the clay-rich MGS-1C would give a negative electric field, with the silica-rich MGS-1 giving  a positive field.Instead, the opposite was seen here.The magnetic signal of this drop again showed nothing beyond the background noise of the lab.

Discussion
An experimental apparatus has been designed and constructed for generating and measuring both electrical and magnetic signals for charged dust samples.The new apparatus both remotely senses electric field from charged granular material and detects the material directly in a Faraday cup.Electrical signals were detected from falling regolith simulant both remotely and directly.The apparatus also remotely senses magnetic fields.Suggestive magnetic fields were detected which are present when the apparatus is used in its rotational mode, but not in the dropping mode.
It was also shown that the MGS-1 regolith simulant charges significantly in a vertical drop, more than than the bimodal glass bead drop.It is also shown that in this fall, an electric field is detected, which lags the drop -this is likely due to the high number percentage of small particles that are aloft for much longer.No magnetic field was observed in the falls, as expected from the lack of rotational motion.
Differences were observed between drops of MGS-1 and MGS-1C.The MGS-1C experiences a much lower net charge than the MGS-1, and the magnitude of the field is lower as a result of this.The two simulants have differing compositions, and this changes the direction of the electric field detected.
Further development is planned to improve the rotational motion set-up such that the glass beads, and regolith simulant can be used.This redesign will combine the concepts of the 'dust fountain' from Forward et al. [19][20] and the experiments regarding vortex formation [21].In essence, the dust fountain approach is planned, with air blowing up from the bottom of the tank -this alone is enough to impart a charge to the particulate, and an upward motion.When combined with air blowing from the sides along the walls of the tank, a significant vortex is formed.Thus, the horizontal mixing mode can be refined -and indeed, the conditions of the formation of a dust devil can be better replicated.There is also a need to test this in a tank designed to better replicate the Martian atmosphere -in low pressure carbon dioxide.
Future work is also planned to develop a better understanding of the expected magnetic fields from Martian dust storms by utilizing data from NASA's Insight lander.To supplement this, more terrestrial magnetic field data from dust storms will be gathered in-situ in an extended fieldwork campaign.

Figure 1 :
Figure 1: Example of a Martian vista from NASA's Mars 2020 Perseverance rover.This is composite from 3 images taken by the left navigation camera on Sol 15.Image credit NASA/JPL-Caltech.

Figure 2 :
Figure 2: The stages of dust devil formation.
(a) The formation of an electric field between two regions of opposing charge within a dust devil, separated by a characteristic distance d.(b)The formation of an magnetic field by cyclic rotation around the central axis of the devil.The characteristic dimension, A, is the cross-sectional area.

Figure 3 :
Figure 3: Illustration of the formation of electric and magnetic fields within a dust devil.The charged particles in the dust devil are represented in black, with red showing the dipole moment and green the field.The key dimension related to the formation of each dipole is shown in purple. [9].

Figure 4 :
Figure 4: Schematic diagram of experimental apparatus in the 2 configurations used.

Figure 5 :
Figure 5: Example waveform from tank capacitance experiment.Drive voltage is shown in blue and current in grey, trended in red.This shows the expected current plateaus for each of the rising and falling limbs.

Figure 6 :
Figure 6: Charge measurement response from drop of glass beads with charge generated in the drop.

Figure 7 :
Figure 7: Electrical measurement response from 6 drops of pre-charged glass beads.The thick black line shows the mean signal from each instrument.

Figure 8 :
Figure 8: Power spectrum (left) and spectrogram (right) for an example of horizontal rotation of polystryene balls.The Earth's background DC component has been subtracted.In the power spectrum, MSD refers to the magnetic spectral density.

Figure 9 :
Figure 9: Charge recorded on the Faraday cup, and electric field traces for a drop of MGS-1 regolith simulant (blue line) and MGS-1C simulant (orange line).

Table 1 :
Summary of voltage rates, capacitive current and calculated capacitance for each experimental run.

Table 2 :
The conversion of sieve mass percentage in bins, to number percentage for the same bins for MGS-1 simulant.This assumes a density of 1.29 gcm -3 [16][17], and spherical particles of a diameter equal to the mid point of each bin.Binned Diameter [µm] Mass Retained Percentage [%] [17] Number Percentage [%]