Simulation of the Martian dust aerosol at low wind speeds

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. E12, 5133, doi: /2001je001807, 2002 Simulation of the Martian dust aerosol at low wind speeds J. P. Merrison, 1 P. Bertelsen, 2 C. Frandsen, 2 P. Gunnlaugsson, 3 J. M. Knudsen, 2 S. Lunt, 1 M. B. Madsen, 2 L. A. Mossin, 4 J. Nielsen, 3 P. Nørnberg, 4 K. R. Rasmussen, 4 and E. Uggerhøj 1 Received 26 October 2001; revised 31 January 2002; accepted 13 June 2002; published 19 December [1] Performing realistic simulations is crucial for developing, testing, and subsequently analyzing results of experiments sent to the surface of Mars. A wind tunnel has been constructed, in which the atmospheric conditions of pressure and wind speed are controlled to match those observed by the Pathfinder mission to Mars. Injection into the wind tunnel of an analogue dust from Salten Skov in Denmark allows simulation of the Martian aerosol. Here experiments can be tested in preparation for a planned mission to the planet (Mars Exploration Rovers to be launched in 2003). Observations of adhesion and cohesion effects have been made in the wind tunnel, which are relevant to particle transport and of significance for validating the performance of specific experiments on Mars. Preliminary studies have been made, at Mars atmospheric pressure, of dust capture on magnet arrays similar to those flown on the Mars Pathfinder mission. INDEX TERMS: 6225 Planetology: Solar System Objects: Mars; 0305 Atmospheric Composition and Structure: Aerosols and particles (0345, 4801); 5470 Planetology: Solid Surface Planets: Surface materials and properties; 0343 Atmospheric Composition and Structure: Planetary atmospheres (5405, 5407, 5409, 5704, 5705, 5707); KEYWORDS: Mars, simulation, aerosol, wind, mineralogy, dust Citation: Merrison, J. P., et al., Simulation of the Martian dust aerosol at low wind speeds, J. Geophys. Res., 107(E12), 5133, doi: /2001je001807, Introduction [2] A wide ranging multinational program of exploration is planned for Mars over the next few years. A variety of experiments are to investigate the Martian atmosphere, suspended dust, and wind properties. It is hoped that the present development of a Martian aerosol simulation facility can prove useful in developing, testing, and subsequently analyzing results of such experiments sent to Mars. Dust is a significant factor affecting the dynamics of the Martian atmosphere since it affects heating rates of the gas in which the dust is suspended and therefore effects atmospheric circulation [Zurek et al., 1992; Edgett and Malin, 2000]. It is also an important environmental factor; ultrasmall suspended particles readily adhere to all types of surfaces causing potential mechanical and electrical hazards [Appelbaum et al., 1993; Landis and Jenkins, 2000]. The most important difference between the atmosphere of Mars and that of Earth is the low hydrostatic pressure on Mars, around 1% that of Earth. The work presented here will concentrate on the effects this low pressure has on the dynamics of the aerosol. The study of adhesion of dust particles from the Martian atmosphere and the process of capture using 1 Institute for Storage Ring Facilities, Aarhus University, Denmark. 2 Ørsted Laboratory, Niels Bohr Institute for Astronomy, Physics and Geophysics, Copenhagen, Denmark. 3 Institute for Physics and Astronomy, Aarhus University, Denmark. 4 Department of Earth Sciences, Aarhus University, Denmark. Copyright 2002 by the American Geophysical Union /02/2001JE magnets will be made; both depend on the dynamics of the wind flow. In the wind tunnel developed at Aarhus University this flow can be controlled to correspond to that observed on Mars, specifically to correspond to the parameters derived from the results from the Mars Pathfinder mission [Schofield et al., 1997]. [3] Simulations of the magnetic properties experiments are to be performed, where magnetic dust particles will be captured using permanent magnets [Madsen et al., 1999; Bell et al., 2000]. These dust capture simulations aim to develop a deeper understanding of the process of capture by reproducing the correct aerodynamic environment. Some of the assumptions used in previous, detailed analysis can be tested [Gunnlaugsson et al., 1998; Hargraves et al., 2000; Gunnlaugsson, 2000]. 2. Wind Tunnel and Atmospheric Simulation Chamber [4] In our experimental arrangement, a recirculating wind tunnel measuring approximately 0.3 m wide, 0.4 m high, and 2 m long is enclosed in a vacuum chamber (see Figure 1) itself of dimensions 3 m long and 1 m in diameter. The chamber can be evacuated to 3 Pa ( mbar) using a rotary roughing pump. The vacuum is monitored using a calibrated pirani gauge. This chamber can be vented either to air or other selected gas mixtures, though the experiments presented in this manuscript have focused on lowpressure air. A motor driven fan can maintain wind speeds of 0 7 m s 1. These wind speeds are far less than those used in previous studies [White, 1979, 1981; White et al., 16-1

2 16-2 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL Figure 1. Showing the Aarhus Mars simulation atmospheric chamber, housing a recirculating wind tunnel (as seen from above). 1997] and the present studies are intended to complement some of this earlier work. Unless otherwise stated, in these studies a wind speed of 0.45 m s 1 and a pressure of 10 mbar were used. As will be discussed below, at higherwind speeds depletion of suspended dust particles due to adhesion on internal walls of the chamber becomes rapid and aerodynamic studies become difficult in the limited time available. The top of the wind tunnel consisted of a large glass plate with a metal frame. Two CCD cameras were mounted here, one of which could be remotely controlled to pan the wind tunnel interior (at low resolution) and the other was fixed to observe specific samples (at higher resolution). Instruments could be observed in situ as they accumulated dust particles. [5] The aerosol is produced by admitting a mixture of gas (usually air) and dust into the wind tunnel through a 3 mm wide copper tube. The jet thus formed reaches high speed exiting the tube. This speed can be estimated knowing the gas pressure and density (assuming adiabatic expansion), a value of around 400 m s 1 is obtained. The air carrying the dust particles causes some increase in the ambient pressure in the chamber during injection. This increase was usually less than 0.5 mbar giving a <5% change at a nominal chamber pressure of 10 mbar. The downstream wind speed in the wind tunnel was measured to increase briefly and become extremely turbulent, varying by ±1.5 m s 1 during injection. This turbulence lasted around 10 s. This method of introducing particles into suspension is similar to that employed in previous studies [White, 1981]. [6] One foreseeable problem in these simulations was that this increased wind speed and turbulence during dust (aerosol) injection could cause an atypical accumulation or loss of captured dust, even though this lasts only a few seconds. This potential problem was investigated using a rotating table, which was housed such that components could be exposed to the wind tunnel aerosol and also, if so selected, enclosed in a sealed compartment. This rotating table could be controlled remotely. Exposure for the first s during dust injection could be compared with the amount captured after exposure when the wind flow had stabilized. Maghemite was used for these tests as well as the Salten Skov analogue. Clear qualitative differences were observed in the deposition of dust. During injection the deposition appeared more symmetric on the magnet array and more symmetric on the tilt magnet (the circle was full) compared with the case in a steady wind when accumulation is almost solely on the strong field section. This is probably due to the changing wind direction during aerosol injection. No removal of dust was observed on performing the injection procedure (without dust). [7] The velocity of individual suspended dust particles was measured using a commercial (Dantec) flow velocity analyzer based on a laser Doppler anemometer system. Here two laser beams are passed through a glass window into the wind tunnel. Within the volume where the beams overlap (situated around 400 cm from the emitting head) a dust particle will periodically reflect laser light due to interference of the beams. The scattered light from this particle will be detected by photomultipliers working in backscattering geometry. The intersection volume of the crossed beams, described by a Gaussian function, is around 0.65 mm wide (falling to 1/e of the maximum intensity). This volume is considered to be that sampled by the device and can be converted into a wind facing cross-sectional area (A = cm 2 ). The axial component of the velocity (i.e., along the wind direction) is measured and the turbulence is determined (given in %) from the width of the velocity distribution. [8] Various suspended mineral powders were used, these included clay, some iron oxides, Salten Skov soil, and also smoke from a vaporizing fog generator (oil vapor aerosol generator). In this latter case extremely small, low-density particles were suspended and measurement of their velocity was assumed to give a good determination of the actual wind speed. Agreement in the measured wind speeds was obtained in all cases, indicating that all the different dust particles studied follow closely the ambient wind speed. [9] Due to the small size of the present wind tunnel and due to the far from optimal circulation system, at higher pressures (>100 mbar) the wind flow is highly turbulent (over 30%). At low pressures, however, due to the decrease in Reynolds numbers and rescaling of the boundary layer, the turbulence in the central 10 cm wide area of the tunnel falls below 5% for the typical wind speeds considered in these studies (i.e., <2.0 m s 1 ). The dependence of turbulence on wind speed is illustrated in Figure 2; the drastic change observed at different gas pressures emphasizes its importance for the correct simulation of wind dynamics. 3. Dynamics of the Gas Flow [10] The atmospheric pressure on Mars is only around 7 mbar (6 12 mbar), this has a significant effect on the forces, flow, and scale of wind induced behavior compared with Earth. As will be seen, small particles (less than 10 mm) can still be kept in suspension even at the lowest of the measured wind speeds on Mars, around 1 m s 1 [Schofield et al., 1997]. However, this is far from the wind speeds required to entrain (lift-off) such particles from rest once they have adhered (or cohered) to a surface [White et al., 1997]. On Mars one process by which dust entrainment is supposed to occur is through dust devils, i.e., warm core vortices that form at the base of convective plumes. Such dust devils have also been observed and are in fact relatively abundant in desert areas on Earth. Evidence for hundreds of such storms has been observed on Mars [Thomas and Gierasch, 1985; Metzger et al., 1999; Renno et al., 2000].

3 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL 16-3 Figure 2. Laser Doppler measurements of wind turbulence at a central cross section of the wind tunnel. An injected Mars simulation aerosol is used at different gas pressures and wind speeds. Another possible mechanism involves the impact of larger saltating particles [Greeley et al., 2000]. [11] Turbulent eddies in the wind stream are responsible for particle lift. The so-called friction velocity of the wind (u*) is used to approximate this vertical component of wind turbulence. If the friction speed exceeds the terminal velocity (U f ) of a dust particle (U f < u*), then the wind is considered capable of suspending that particle. In this simulation the friction velocity was determined by studying the wind flow around a large flat plate, which was introduced close to the center of the wind tunnel. Wind speed measurements were obtained as a function of distance from the plate, called the z direction. Ignoring the viscous sublayer, which is present up to around 1 cm from the plate, a logarithmic dependence is expected. This is according to the so-called law of the wall in which the gradient of this logarithmic dependence should be constant and yield a value of the friction velocity (u*) by the expression: du/dz = 2.4 ln(u*) [Monin and Yaglom, 1965; Coles, 1962]. Figure 3 shows a typical data set obtained at 10 mbar for a (free stream) wind speed of around 0.45 m s 1.Inthis case u* =3.3cms 1 was obtained by fitting. [12] The force exerted by the wind turbulence on a dust particle can be calculated using F = C d ru * 2 pd 2 /8, where C d is the drag coefficient, r is the gas density, and d is the particle diameter [Greeley and Iversen, 1985]. At low pressures the medium is well described by an ideal gas and Stokes law is considered accurate, hence the drag coefficient can be obtained knowing the Reynolds number (Re*); C d = 24/Re*. The Reynolds number is a dimensionless parameter commonly used when discussing turbulent fluids. In the case of a dust particle with diameter d, the Reynolds number is given by: Re* = u*d/s, where the kinematic viscosity (s) is merely the standard molecular viscosity (m) divided by the density of the medium (i.e., s = m/r). At 10 mbar air pressure and T = 293 K, as used in the simulations, s = 14.6 cm 2 s 1. However, in this low-density gas the suspended dust particles are significantly smaller than the collisional mean free path (l = 6.6 mm). The Figure 3. Law of the wall measurements. Velocity profile measured from a plate parallel to the wind direction showing the logarithmic dependence from which the friction velocity can be calculated u * =3.26cms 1. Free flow wind speed u =45cms 1. Pressure = 10 mbar. medium can therefore no longer be considered continuous and a correction must be applied. The so-called slip factor reduces the effective velocity of the wind with respect to the suspended particles. The slip factor is given by the expression: f a =1+(2l/d)[ exp( 1.57d/2l)] [Davies, 1945] and in the case of spherical particles with d =1.5mm a value of f a = 10.8 is obtained. The corrected Reynolds number becomes Re* = This gives a value for the drag coefficient of C d = (see Table 1). [13] From these values, the terminal velocity in the simulation atmosphere can be calculated by equating the forces of drag and that of gravity, U f = r p gd 2 /18rs, where r p is the density of the dust particle. After correction for slip this gives U f =0.38cms 1. This is a factor of 10 smaller than u*, and the particles can therefore be considered well suspended. For larger particles, the slip correction is not necessary. The largest (spherical) particles, which may be kept in suspension, can be calculated, taking U f = u*. This yields a value of around 16 mm again for a wind speed of 0.45 m s Mars Simulation Dust [14] By dust we refer to particles of less than 63 mm in diameter, which have been obtained by sieving. The Salten Table 1. Aerodynamic Parameters Used in the Simulation a Parameter Value Gas density (r) gcm 3 Kinematic viscocity (s) 14.6 cm 2 s 1 Mean free path (l) 6.6 mm Friction velocity (wind speed = 45 cm s 1 ) 3.3 cm s 1 Terminal velocity (U f ) b 0.38 cm s 1 Knudsen number (Kn = l/d) b 4.4 Reynolds number (Re) b Drag coefficient (C d ) b a Air pressure = 10 mbar; temperature = 293 K. b Particle diameter d =1.5mm; particle density r p = 5.2 g cm 3.

4 16-4 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL Figure 4. Laser scattering technique measuring the particle size distribution (by volume) for samples of Salten Skov obtained from: (left) wind tunnel wall and (right) bottom of the wind tunnel. Skov Martian dust analogue is a chemical sediment precipitated from iron II bearing ground water. The sieved fraction contains approximately 60% iron by weight and the mineralogical composition is 73% goethite, 14% hematite, and 13% maghemite as determined by Mössbauer spectroscopy. The most important aspect of the Salten Skov Mars analogue for the present studies is its magnetic properties, which are intended to resemble those of the Martian dust. One aim of these simulation experiments is to demonstrate this similarity quantitatively. Comparison is made with samples of pure minerals (iron oxides): hematite (a- Fe 2 O 3 ), maghemite (g-fe 2 O 3 ), and magnetite (Fe 3 O 4 ). Hematite is only weakly magnetic, typically the magnetic susceptibility is around c lf = m 3 kg 1, whereas maghemite and magnetite are strongly magnetic, with susceptibility around c lf = m 3 kg 1 [Morris et al., 2000]. [15] To obtain reproducible results, dust was measured out by volume in units of around 0.1 cm 3. In the case of Salten Skov dust this weighed around 0.12 g. The number of particles subsequently suspended upon injection of this quantity of dust, however, varied by around a factor of 2, as measured by the laser Doppler anemometer. As discussed these powders readily form aggregates. Also, as evaluated by visual inspection, a substantial component of the powder consisted of larger aggregates/particles. The particle size distribution would be expected to play an important role in suspension. Interestingly weighing the powders, which had been measured by volume, showed that the amount of mass they contained varied by a factor of 2. For the minerals magnetite, hematite, and maghemite, which have similar molecular weight, the volume of dust contained was 0.22, 0.11, and 0.11g, respectively. Clearly the packing of these powders is different. A clay was used, here one measure also weighed around 0.11 g, despite the bulk density of this mineral being only around 2.7 g cm 3 compared with around 5.2 g cm 3 for the iron oxide minerals. [16] A laser diffraction technique (Low Angle Laser Light Scattering) was used to determine the size distribution of the dust particles. This technique was sensitive down to around 0.5 mm. Measurements were taken both before injection and of samples obtained after the experiment from surfaces within the wind tunnel. The dust was suspended in water for this analysis and ultrasonically agitated. Break up of aggregates was anticipated during this process and indeed after extended exposure to ultrasound a distinct shift to smaller particles (2 mm) could be seen. However, during the measurements it was observed that aggregates might also spontaneously form in the water suspension. In this particle size analysis a volume distribution is obtained, which is equivalent to a weight distribution assuming that the density of the particles is uniform. Compared with number density this enhances the distribution of larger particles. [17] The total number of dust particles injected into the chamber (N i ) can be crudely estimated from the injected mass (M) of material, assuming each particle has the average measured diameter (spherical volume) and density of the bulk material r p ; N i =3M/4pr p (d/2) 3. Typical values were in the range g 1 (scaled to M = 1 g). The variation, however was great, the lowest value was under Dust Adhesion and Cohesion [18] Particle size distribution by volume was measured with the laser diffraction technique using Salten dust samples, which had adhered to the wall of the wind tunnel and from deposits at the bottom of the wind tunnel (see Figure 4). There appears to be two distinct distributions present: one of roughly 2 mm diameter and another with a size distribution centered around 30 mm (presumably consisting of aggregates of the smaller dust particles). Those that have accumulated at the bottom of the wind tunnel are rich in much larger particles, which have presumably fallen out of suspension. This is not unexpected at the relatively low wind speeds used in these investigations. [19] Electron microscope (SEM) pictures were also taken of a dust exposed surface, for the various minerals used. The shape of the primary dust particles could in this way be (qualitatively) determined. These are listed in Table 2. It was noted that (for all minerals studied) the majority of the observed dust grains consisted of aggregates, ranging in size from several hundred microns to less than 1 mm (containing only a few primary dust particles). There was also a significant population of what appeared to be primary dust particles with diameters of a few 100 nm (see Figure 5). Most dust grains had a diameter greater than 0.5 mm. [20] As has been demonstrated cohesion (aggregation) of dust clearly occurs in this simulation, either before or after leaving suspension. Large grains have been demonstrated to form, approaching 1mm in diameter. These are probably large enough to be entrained by the wind and could play a role in dust circulation. The dust could alternately leave suspension forming large aggregated grains and then be entrained

5 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL 16-5 Table 2. Dust Grain Properties and Suspension Parameters a Average Particle Diameter by Volume, mm Observed Primary Dust Particle Morphology Number of Particles: (N s /N i ) Suspended/Injected Suspension Time t seconds Single Exponential) Mineral Dust b a-fe 2 O ± 0.2 irregular spheroid b g-fe 2 O ± 0.2 needle b Fe 3 O ± 0.3 irregular spheroid Clay b 8.0 ± 0.5 plate (sheet) Salten fine 2.0 ± 0.2 needle + spheroid Salten aggregated 37 ± 5 needle + spheroid +101, a Average particle diameter (by volume) is measured by laser diffraction. b For the samples a-fe 2 O 3, g-fe 2 O 3,Fe 3 O 4, and clay only the small component is taken for the average particle diameter. For the samples Salten-fine and Salten-aggregated the average diameter of all particles was taken. and reenter suspension. The size, form, and structure of these aggregates will affect their transport. The presence of aggregates on Mars is currently an issue of discussion [Greeley et al., 2000]. Such effects may be investigated in further detail in the Aarhus Mars simulator facility. 6. Dust Suspension [21] While running at Earth atmospheric pressure (1000 mbar) suspension times of several hours were usually observed. At lower pressures this reduces to the order of minutes or less. The suspension time of hematite was measured as a function of free flow wind speed at an atmospheric pressure of 10 mbar. Good agreement is found with an inverse relation between suspension time and wind speed as shown in Figure 6. This indicates that the dust is lost predominantly by adhesion, since at higher-wind speeds the dust grains encounter an adhesive surface more rapidly. At the lowest wind speeds (0.1 m s 1 ) a reduction in suspension time is observed presumably due to loss by gravitation (falling out of suspension). [22] To quantify adhesion or capture of particles one must determine the number of particles that strike a surface (area) during a certain period of time. To calculate this dose of particles the number density within the wind tunnel must be determined. In principle, this can be obtained knowing the mass of material and the distribution of particle sizes. As discussed, these unfortunately cannot be controlled accurately due to cohesion of the particles. The laser Doppler anemometer measurements determine in an independent way the number density in the aerosol. From these measurements the dose may be calculated. The number of particles striking a (wind facing) surface is the product of the suspended particle number density (n), wind speed, and time (dose = ntu). However, since in our low wind speed tests the suspension time of the dust is proportional to 1/u, the dose simplifies to being purely a function of the initial (injected) dust number density. By counting the total number of particles detected by the anemometer (n s ) one obtains a direct measure of the dose received by a surface during the exposure, knowing the detection area (A) of the laser Doppler anemometer; dose = n s /A. For Salten Skov dust n s was typically around counts when injecting around 0.12 g (0.1 cm 3 ) of dust. This corresponds to a dose of cm 2 particles impinging. The suspension time was typically around 118 s. [23] The total number of suspended dust particles inside the simulation chamber (N s ) can also be calculated using N s = dose V/(t u), where V is the internal volume of the chamber and t is the suspension time. Values were usually in the range g 1 of injected material, varying by little more than a factor of 10. This number can be compared with the total number of dust particles injected into the chamber (N i ), as estimated previously. Values of the ratio N s /N i are compiled in Table 2. In all cases this ratio is significantly greater than 1, indicating that substantial particle breakup is occurring on injection, presumably due to the high turbulent wind speeds at the nossle (as discussed). This is supported by the observation that the Figure 5. An electron microscope (SEM) image of maghemite grains prior to injection. Figure 6. The suspension time of hematite dust at 11 mbar pressure and various wind speeds showing a 1/u dependence.

6 16-6 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL smallest increase in particle number is seen for the finely sorted Salten dust sample and by far the greatest increase was found with the aggregated Salten dust sample. It seems that even the small component of the measured dust size distribution undergoes breakup and must therefore consist of a large fraction of aggregates. Morphology appears to play an important role in cohesion with spheroidal, needle, and sheet formed particles, respectively, showing increasing degrees of breakup on injection. These forms have, respectively, increasing surface area, which would be expected to enhance aggregation. [24] The suspension time of pure minerals could be fitted well with a single exponential. No improvement was found with the addition of multiple-exponential components. For smaller particles the adhesion force, which is a function of the surface area, will be greater compared with the particle momentum, which is a function of the volume. A higher sticking coefficient would therefore be expected for smaller particles. If particle loss is dominated by adhesion (as observed) they would be lost more rapidly and have a shorter lifetime than larger particles. Given this difference in adhesion rate for different particle sizes the singleexponential component is surprising and may indicate that only a narrow range of dust sizes are suspended. The suspension times also only vary by less than a factor of 2 (Table 2), again indicating a similarity in suspended particle size. The fine Salten dust is seen to have a slightly shorter suspension time than the aggregated Salten sample, as would be expected from the above argument. [25] Comparing the average injected particle diameter of the four pure minerals with suspension time (see Table 2) it appears that with smaller particles an increase in suspension time occurs. However, as discussed above it seems that this measured injected particle diameter (using the laser diffraction technique) is not suitable for evaluating suspended particle size due to aggregate formation and breakup. In addition to particle size several other factors must also be considered that affect suspension time. Morphology seems to play an important role (as discussed) since increased surface area would be expected to enhance adhesion and reduce suspension time. Reduced bulk density of the dust would be expected to enhance adhesion by reducing particle momentum, though with the exception of clay this is not relevant to these studies. The strength of the adhesive force (whether electrochemical or electrostatic) is probably material-dependent, this factor adds another degree of complexity. It is not at present possible to separate all of these factors. 7. Magnetic Capture [26] Using this Mars simulation wind tunnel it is possible to address some of the aerodynamic issues related to the capture of magnetic dust from a low-density aerosol. Such work is required in order to expand the knowledge based on the previous empirical and analytical work on capture cross sections from the Pathfinder mission [Gunnlaugsson et al., 1998; Gunnlaugsson, 2000]. Specifically some of the assumptions used can be tested. It is also hoped to demonstrate the usefulness of the simulation facility for testing new magnet designs to be sent to Mars in Figure 7. Tip plate magnets having collected suspended maghemite dust. (left) Wind from the forward direction (0.45 m s 1 ) and (right) the same wind speed with the sample continually rotated (random wind direction). Compared with above: the Mars Pathfinder Tip Plate Magnet having collected Mars dust for 78 sols. [27] An important aspect of magnetic particle capture could be the orientation of the magnets with respect to the direction of the wind. The so-called tip plate magnet was therefore exposed to the same dose of dust while stationary and compared with the case where it was rotated i.e., exposed to a direction averaged wind. Here a qualitative difference is observed between the two cases. With steady wind from the forward direction, accumulation of dust is exclusively observed on the strong field section. A wind from random directions causes a closed circle, with deposits also on the weak field section (see Figure 7). This randomwind sample resembles more closely the observations of accumulation on the Mars Pathfinder, which used the same design of tip plate magnet [Gunnlaugsson et al., 1998]. The geometry of the surroundings is also important for dust deposition, since flow is affected by obstructing objects. Greatly reduced dust capture was observed when the magnet array was placed close (around cm) to even relatively small obstructions, little larger than the array itself. If the obstruction is upwind and close to the magnetic array no visible accumulation of Salten Skov dust was observed after a dose of particles. [28] During the Pathfinder mission similar low wind speeds to those in this simulation were observed. Particle densities from the Viking mission were determined to be around 1 10 cm 3 [Pollack et al., 1995]. With this information it is possible to compare the capture efficiency of magnets in the simulation to those observed after a similar dose exposure on Mars. Assuming a wind speed of 1 m s 1 and a particle density of 1cm 3 one obtains for 1 sol dose From the Pathfinder mission a visible pattern was clearly visible on the strongest magnet after a dose of a few 10 7 cm 2 (a few sols) and a pattern was observed on the next strongest magnet of the array after around (10 20 sols). In the wind tunnel experiments (using Salten Skov dust) a pattern was discernible both on the first and second magnets of the array after a dose of only

7 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL 16-7 cm 2, indicating an accumulation rate around 50 times greater than that on Mars. The reason for this discrepancy is not as yet clear, though there are several possibilities. The most obvious explanation is that the Salten Skov dust material has different magnetic properties than the suspended dust on Mars, i.e., being more strongly magnetic than the Mars dust. However, there are other possible explanations for the dose disagreement. One is the effect of wind direction or obstruction on Mars, as discussed. Also, the flow in the simulation is far from ideal and undoubtedly more turbulent than the free flow wind stream on Mars, this could also play a role in capture. Particle size is of importance in the interpretation of accumulation rates observed on Mars and in principle could explain some of the disagreement [Gunnlaugsson, 2000]. Larger particles are more visible, giving an enhancement in the observed pattern intensity; they are also expected to be easier to attract out of suspension. The magnetic force acting is (presumably) a bulk effect and proportional to the volume of the particle, whereas suspension forces affect the surface. However, as discussed, the suspended dust in the simulation appears to be finely divided (with average diameter less than 2.2 mm) and not significantly larger than the suspended dust observed on Mars (around 1.6 mm). [29] Microscopic observations of the accumulated particles on the magnets have shown that they have formed aggregates, some of them of order 100 mm. These clearly form after capture since particles of such size cannot be suspended at these wind speeds. Minerals containing a strong magnetic component were observed to form vertical structures (chains), accumulating along field lines. The presence of such structures might be used as an indicator for the presence of a strongly magnetic component. [30] It has been difficult to perform quantitative comparison of dust accumulation remotely using the CCD cameras. Optical band pass filters have not been used here, as was the case on the Pathfinder mission. Proper lighting conditions and a well-selected camera observation angle are vital for obtaining good contrast in the images. The dust particles can appear as light or dark compared to the background platinum and for certain angles the pattern therefore disappears. Similarly on Mars the variation in position of the optical source (the sun) in relation to the magnets is a problem for quantitative analysis of dust accumulation. In these simulations, where it was not possible to arrange reproducible lighting and camera angle, it was necessary to remove the samples and view them with a hand held camera. The application of artificial lighting would be advantageous on Mars. 8. Conclusions [31] A low-pressure wind tunnel has been constructed in order to simulate the Martian aerosol. It has been demonstrated to produce a well-characterized flow, in the pressure ranges found on Mars and at wind speeds similar to those observed by pathfinder (0 10 m s 1 ). Adhesion and cohesion effects of the dust have been seen to be crucial in the interpretation and operation of experiments in this environment. Experiments intended for the NASA 2003 Mars mission are to be tested and modeled in this Mars simulation wind tunnel. It has been shown that capture patterns can be reproduced as observed in the magnetic properties experiments on Mars. Differences in capture rate are observed between Mars Pathfinder observations and our simulation using Salten Skov analogue dust. It will require a more thorough study to fully understand these differences, possibly modeling of the Mars lander. Aggregation causes great variation in the density of the dust when in powder form. The number that becomes suspended is more constant however, this study indicated that the average suspended particle diameter was similar for all samples, being in the range mm. This is close to the particle size observed on Mars. In future work, it would be desirable to measure in situ the size and shape of the suspended particles. This might resolve some of the problems of cohesion/adhesion and lead to a better understanding of the process of magnetic capture. The current work has concentrated on the low-pressure behavior of a simulated Martian aerosol, however the experimental arrangement is to be modified in order to achieve low temperatures (around 200 K) as on Mars. In future, the atmospheric gas mixture will also be controlled in order to reproduce the atmosphere on Mars more accurately. [32] Acknowledgments. We acknowledge Joanna Abraham for photographic help, J. D. Iversen and R. B. Hargraves for useful discussion and the financial support of the Ib Henriksens Fond and ESA-Følgeforskning. References Appelbaum, J., G. A. Landis, and I. Sherman, Solar energy on Mars: Stationary collectors, J. Propul. Power, 11, 554, Bell, J. F., III, et al., Mineralogic and compositional properties of Martian soil and dust: Results from Mars Pathfinder, J. Geophys. Res., 105, 1721, Coles, D. E., The turbulent boundary layer in a compressible fluid, Rand Corp. Rep., 403-PR, Davies, C. N., Definitive equations for the fluid resistance of spheres, Proc. Phys. Soc., 57, 18, Edgett, K. S., and M. C. Malin, New views of Mars eolian activity, materials, and surface properties: Three vignettes from the Mars Global Surveyor Mars Orbit Camera, J. Geophys. Res., 105, 1623, Greeley, R., and J. D. Iversen, Wind as a Geological Process, Cambridge Univ. Press, New York, Greeley, R., M. D. Kraft, R. O. Kuzmin, and N. T. Bridges, Mars Pathfinder landing site: Evidence for a change in wind regime from lander and orbiter data, J. Geophys. Res., 105, 1829, Gunnlaugsson, H. P., Analysis of the magnetic properties experiment data on Mars: Results from Mars Pathfinder, Planet. Space Sci., 48, 1491, Gunnlaugsson, H. P., S. F. Hviid, J. M. Knudsen, and M. B. Madsen, Instruments for the magnetic properties experiments on Mars Pathfinder, Planet. Space Sci., 46, 449, Hargraves, R. B., J. M. Knudsen, P. Bertelsen, W. Goetz, H. P. Gunnlaugsson, S. F. Hviid, M. B. Madsen, and M. Olsen, Magnetic enhancement on the surface of Mars, J. Geophys. Res., 105, 1819, Landis, G. A., and P. P. Jenkins, Measurement of the settling rate of atmospheric dust on Mars by the MAE instrument on Mars Pathfinder, J. Geophys. Res., 105, 1855, Madsen, M. B., R. B. Hargraves, S. F. Hviid, H. P. Gunnlausson, J. M. Knudsen, W. Goetz, C. T. Pedersen, A. R. Dinesen, C. T. Mogensen, and M. Olsen, The magnetic properties experiments on Mars Pathfinder, J. Geophys. Res., 104, 8761, Metzger, S. M., J. R. Carr, J. R. Johnson, T. J. Parker, and M. T. Lemmon, Dust devil vortices seen by the Mars Pathfinder camera, Geophys. Res. Lett., 26, 2781, Monin, A. S., and A. M. Yaglom, Statistical Fluid Mechanics, p. 257, Nauka, Moscow, Morris, R. V., et al., Mineralogical, composition, and alteration of Mars pathfinder rocks and soils: Evidence from multispectral, elemental, and magnetic data on terrestrial analogue, SNC meteorite, and Pathfinder samples, J. Geophys. Res., 105, 1757, Pollack, J. B., M. E. Ockert-Bell, and M. K. Shepard, Viking lander image analysis of Martian atmospheric dust, J. Geophys. Res., 100, 5235, 1995.

8 16-8 MERRISON ET AL.: SIMULATION OF THE MARTIAN DUST AEROSOL Renno, N. O., A. A. Nash, J. Lunine, and J. Murphy, Martian and terrestrial dust devils: Test of a scaling theory using Pathfinder data, J. Geophys. Res., 105, 1859, Schofield, J. T., J. R. Barnes, D. Crisp, R. M. Haberle, S. Larsen, J. A. Magalhães, J. R. Murphy, S. Seiff, and G. Wilson, The Mars Pathfinder Atmospheric Structure Investigation/Meteorology (ASI/Met) experiment, Science, 278, 1752, Thomas,, and Gierasch, Dust devils on Mars, Science, 230, 175, White, B. R., Low Reynolds number turbulent boundary layers, in Turbulent Boundary Layers: Forced, Incompressible, Non-reacting, edited by H. E. Weber, p. 209, Am. Soc. Mech. Eng., New York, White, B. R., Low-Reynolds-number turbulent boundary layers, J. Fluid Eng., 103, 624, White, B. R., R. Greeley, B. M. Lacchia, and R. N. Leach, Aeolian behaviour of dust in a simulated Martian environment, J. Geophys. Res., 102, 25,629, Zurek, R. W., J. R. Barnes, R. M. Haberle, J. B. Pollack, J. E. Tillman, and C. B. Leovy, Dynamics of the atmosphere of Mars, in Mars, edited by H. H. Kieffer et al., pp , Univ. of Ariz. Press, Tucson, P. Bertelsen, C. Frandsen, J. M. Knudsen, and M. B. Madsen, Ørsted Laboratory, Niels Bohr Institute for Astronomy, Physics and Geophysics, Copenhagen, Denmark. P. Gunnlaugsson and J. Nielsen, Institute for Physics and Astronomy, Aarhus University, Denmark. S. Lunt, J. P. Merrison, and E. Uggerhøj, Institute Storage Ring Facilities, Physics and Astronomy, Aarhus University, Ny Munkegade, Aarhus, 8000c, Denmark. (merrison@ifa.au.dk) L. A. Mossin, P. Nørnberg, and K. R. Rasmussen, Department of Earth Sciences, Aarhus University, Denmark.