Chondrules born in plasma? Simulation of gas–grain interaction using plasma arcs with applications to chondrule and cosmic spherule formation


  • A. MORLOK,

    Corresponding author
    1. Department of Physical Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
    2. Current address: Institut für Planetologie, Wilhelm-Klemm Strasse 10, 48149 Münster, Germany
    Search for more papers by this author
  • Y. C. SUTTON,

    1. Department of Physical Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
    Search for more papers by this author

    1. Department of Physical Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
    Search for more papers by this author
  • Monica M. GRADY

    1. Department of Physical Sciences, The Open University, Walton Hall, Milton Keynes MK7 6AA, UK
    2. Department of Mineralogy, The Natural History Museum, Cromwell Road, London SW7 5BD, UK
    Search for more papers by this author

*Corresponding author. E-mail:


We are investigating chondrule formation by nebular shock waves, using hot plasma as an analog of the heated gas produced by a shock wave as it passes through the protoplanetary environment. Precursor material (mainly silicates, plus metal, and sulfide) was dropped through the plasma in a basic experimental set-up designed to simulate gas–grain collisions in an unconstrained spatial environment (i.e., no interaction with furnace walls during formation). These experiments were undertaken in air (at atmospheric pressure), to act as a “proof-of-principle”—could chondrules, or chondrule-analog objects (CAO), be formed by gas–grain interaction initiated by shock fronts? Our results showed that if accelerating material through a fixed plasma field is a valid simulation of a supersonic shock wave traveling through a cloud of gas and dust, then CAO certainly could be formed by this process. Melting of and mixing between starting materials occurred, indicating temperatures of at least 1266 °C (the olivine-feldspar eutectic). The production of CAO with mixed mineralogy from monomineralic starting materials also shows that collisions between particles are an important mechanism within the chondrule formation process, such that dust aggregates are not necessarily required as chondrule precursors. Not surprisingly, there were significant differences between the synthetic CAO and natural chondrules, presumably mainly because of the oxidizing conditions of the experiment. Results also show similarity to features of micrometeorites like cosmic spherules, particularly the dendritic pattern of iron oxide crystallites produced on micrometeorites by oxidation during atmospheric entry and the formation of vesicles by evaporation of sulfides.


Chondrules are spherical to sub-spherical silicate-rich objects that range in size from tens of microns to millimeters. They are the dominant component in primitive chondritic meteorites, with modal abundances up to 80 vol%, and have ancient ages, around 4.57 Gyr (Brearley and Jones 1998; Zanda 2004). Their ages show that chondrules formed very early in solar system history, and, along with their high abundance, indicate that the formation process(es) of chondrules was a central step in the evolution of the protoplanetary disk. The origin of chondrules is a matter of vigorous and active debate: their shape and internal texture indicate that they formed in a high-temperature process, such that they are quenched droplets of once-molten silicates. As a result, chondrule precursor materials are difficult to identify with certainty. Based on astronomical observations of protoplanetary and debris disks, probable candidates are amorphous condensates or crystalline materials in “dust-balls” of sub-micron sized particles (e.g., Pontoppidan and Brearley 2010). In addition, unmelted relict grains (usually olivine) within chondrules suggest grain sizes of the precursors up to approximately 100 μm (Hewins et al. 2005; Lauretta et al. 2006). Proposed chondrule formation localities fall into one of two categories: either a protoplanetary nebular environment or in a planetary environment (Desch et al. 2005; Sanders and Taylor 2005; Hezel and Palme 2007).

Proposals for chondrule formation mechanisms are even more diverse than their proposed formation locality. Several formation mechanisms are currently favored: in the outflow from the young Sun (the X-wind model; Shu et al. 1997) or in shock waves (Boss and Durisen 2005), but also the involvement of planetary formation processes (Libourel and Krot 2007; Libourel and Chaussidon 2011). The purpose of this study was to investigate chondrule formation in the context of the latter group of theories. Chondrule compositions and texture are significant indicators of formation process, and there have been several experimental studies designed to investigate chondrule formation. One sequence of studies was made using furnaces for slow heating (e.g., Hewins and Connolly 1996) or flash heating (e.g., Connolly et al. 1998). Results gave an outline for the thermal history of chondrules, including heating time, peak temperature reached, and cooling rates, showing that most chondrules were heated to 1500–1850 °C, followed by a rapid cooling between 10 and up to 3000 °C hr−1 (Desch and Connolly 2002). The experiments successfully replicated chondrule textures and reproduced gas/melt interactions (see Lofgren 1996; Hewins et al. 2005), but were not necessarily a good reflection of a circumstellar nebular environment of formation. This is because, whilst chondrules formed in a free-floating environment, furnace-based simulations necessarily constrain the space in which the starting materials react, and allow the possibility of reactions with furnace walls, etc., thus affecting cooling rate.

In a different series of experiments, chondrule formation by lightning was tested using plasma discharges (Horányi 1997; Hewins et al. 2000; Güttler et al. 2008). These studies, in which material was placed between two electrodes, were unsuccessful, in that material either evaporated, leaving behind a sintered residue, or a surface layer of material melted, with the remainder unaltered, depending on the energy of the plasma. Indeed, in many cases, the material completely exploded (Güttler et al. 2008). Our set up involved a plasma, but differed in several details from the study of Güttler et al. (2008), the most significant of which was that in their set-up, material remained stationary, whilst in this study, material passed rapidly through the plasma. The implication of this difference in methodology is discussed in the next section.

The aim of our study was to simulate formation of chondrules by gas–grain interaction (friction and collisions) in the nebula. We employed a plasma arc to represent the unconstrained spatial environment of the nebula; this eliminated potential effects of interaction with furnace walls. Here, we report the results of our first series of experiments to produce chondrule-analog objects (CAO), which we undertook as a “proof-of-principle”; we do not, at this stage, attempt to reproduce a specific temperature path for the heated particles. CAO are defined as spherical droplets of materials that have been (partially) processed in high-temperature events.

A second area of application for these experiments is the simulation of the atmospheric re-entry of micrometeorites (MMs). These particles of asteroidal or cometary origin with sizes smaller than approximately 2 mm undergo changes in chemistry and mineralogy while heated during the entry into earth’s atmosphere, when they are decelerated from kilometers per second to centimeters per second. As MMs provide an additional source of extraterrestrial material, it is of importance to study the effects of heat during entry on the original mineralogical and chemical composition (Toppani et al. 2001; Genge et al. 2008; Taylor et al. 2011).

Materials and Methods

A basic set-up was designed in which precursor material (mainly silicates) passed through hot plasma created by a plasma arc (Fig. 1). This plasma is intended to be the analog of the heated gas produced by a shock wave as it passes through the protoplanetary environment (Fig. 2). The experiments were undertaken in air (at atmospheric pressure). An atmospheric pressure plasma arc is generated using a low voltage, solid-state switching circuit to drive a Tesla coil (Fig. 1). The coil has a natural frequency in the region of 325 kHz. Through mutual induction, the voltage on the primary electrodes (50 V) is stepped up to achieve a sufficient voltage drop (>20 kV) between the electrodes at the top of the secondary coil, causing breakdown of the air and thus generating the plasma. We also tested the coil under low pressure, and demonstrated that plasma could be generated at pressures down to approximately 10−4 mbar; future investigations will use an inert atmosphere (N2).

Figure 1.

 The experimental setup. Powdered material was dropped through the plasma arc between the two electrodes (see enlarged image on bottom). Run products were collected on a plate set approximately 10 cm below the arc.

Figure 2.

 Schematic interpretation of (a) a shock wave passing through nebula gas to form chondrules and (b) dust grains falling through a plasma arc. Although there are orders of magnitude differences in scale between the nebula and the plasma arc, at the local level of gas-grain interaction, the scale distances are comparable. (c) Variation of grain temperature with time during (b). The width of the plasma arc/distance between arc and collector are shown by the blue lines.

The coil was placed on its side and material dropped vertically through the flame. When ignited, the plasma discharge is rooted at the brass electrodes by visible hot spots, and has a flame-like luminous central column some 5 mm in diameter. The “bowing” of the plasma results from natural convection of the high temperature plasma in the ambient air (hence the term “arc”). This particular configuration, using radio frequency excitation, is able to maintain nonthermal equilibrium conditions with the gas temperature significantly lower than the electron temperature. The gas temperature in the RF plasma is derived directly from the electrical supply without the need for exothermic gas phase chemical reactions.

Generally, in molecular nitrogen-rich plasma, the dominant gas heating mechanism occurs through electron excitation of the molecules’ vibrational states followed by relaxation after further collisions between the now vibrationally excited molecule and the translational state (Kunhardt 2000). Additionally, the rotational states in molecular nitrogen provide a means of measuring the gas temperature. Due to fast relaxation between the rotational and translational states, the rotational temperature can be taken to represent the kinetic gas temperature (Packan et al. 2001). The rotational temperature was determined through fitting of modeled spectra to the observed emission spectrum of the nitrogen second positive system. The wavelength range used for analysis was 362–382 nm, and corresponds to the 0–2 vibrational band transition (Laux et al. 2003). To model the spectrum, a Boltzmann distribution was assumed for the upper state population with the line function of the monochromator presumed to be the dominant broadening mechanism of the spectral lines. The voltage across the plasma was measured using a Tektronix 6015A high voltage probe. Using Ohm’s law, the total current was calculated from the voltage drop across a 100 ohm resistor in series with the top electrode and the ground using a standard oscilloscope probe. The total current consists of both displacement and conduction current components. The conduction current was obtained by making a point-by-point subtraction of the displacement component from the total current. The time-averaged power dissipated in the plasma was calculated from P = VI where V and I are the rms voltage and current, respectively. For a power dissipation of 25 W in the plasma, the corresponding gas temperature is in the region of 2700 °C ± 200 °C.

The experimental set-up differs significantly from that employed by Güttler et al. (2008): in their set-up, material was stationary within the plasma, while in our apparatus, material remained in the plasma for only a fraction of a second. They inferred a maximum temperature of 6500 K during a 60 μs pulse of between 100 and 500 J (Güttler et al. 2008). This contrasts with the experimental conditions of this study, where the grains had a minimum residence time in the plasma of 20 ms at a maximum temperature of approximately 3000 K. (The exact residence time for each grain is unknown—some would fall almost straight through, while others were carried by turbulence and had a slightly extended path-length through the plasma.)

The starting material for the experiments was a mixture of terrestrial minerals of known composition. Na-rich feldspar (Ab76) and Mg-rich olivine (Fo93) were ground, then coarsely sieved, with the 50–100 μm fraction taken for the experiments. The grain size was selected because Hewins and Fox (2004) demonstrated that grain sizes of >20 μm were necessary to produce porphyritic chondrules. Synthetic FeS and Fe metal (both powders from Sigma Aldrich, <1 μm in grain size) were mixed with the silicates. The ratio by weight was 6:4:1:1 of feldspar:olivine:Fe:FeS. In natural chondrules, the feldspar component is usually much lower than that of the ferromagnesian silicates (e.g., Cohen et al. 2004), but since part of the aim of the study was to investigate the general behavior of the different components, we increased the feldspar to track the mesostasis component during crystallization.

The mixture of starting materials, then, was of euhedral to subhedral individual silicate grains covered with the much smaller grains of metal and sulfide powder. Although it was intended that the silicates entered the plasma as a stream of discrete grains, because they were simply dropped manually between the two electrodes, it was likely that some small aggregates formed through electrostatic attraction.

The run products were captured on silica gel pads approximately 10 cm below the plasma arc. The particles spent approximately 0.02 s inside the arc (diameter approximately 5 mm), and dropped onto the collector after approximately 0.15 s. The lack of a high-speed camera precluded accurate measurement of these time intervals, but simple Newtonian dynamics, with the particles falling from rest under gravity onto the collectors set 10 cm under the plasma, yield these approximate values. The results presented are from five runs under the same conditions.

Recognizable spherules were readily identifiable under a low magnification binocular microscope. Larger spherules (>50 μm) were picked and placed on sticky tape on a SEM sample holder and coated with carbon. Images of unprocessed bulk spherules were obtained using a dual beam FEI Quanta 200 SEM at an acceleration voltage of 20 kV and 3 μA beam current in secondary and backscattered electron mode (Fig. 3). Following this, the grains were embedded in resin, polished, and analyzed using the FEI Quanta 200 SEM (fitted with an Oxford Instruments EDX system running INCA software) to determine their composition (Tables 1 and 2; Figs. 4–6).

Figure 3.

 SEM images of complete chondrule-analog objects (CAO) produced in the experiment. (a) Shows anhedral feldspar grains, (b) relict olivine grains in mesostasis. Uncoated and unpolished CAO were placed on SEM stubs and images acquired using variable pressure mode. (c) Is a higher magnification image of the FeO dendrites in (b). (d) Is an example of two CAO fused together similar to compound chondrules. (e) Shows a spherule dominated by FeO-rich dendrites. Large bubbles or vesicles are shown in (f).

Table 1. Average composition (atom %) of olivine in the starting materials and in CAO (Parentheses: Number of analyses N).
rims (72)
  1. – = not detected.

Table 2. Average composition (atom %) of feldspar in the starting materials and in CAO (Parentheses: number of analyses N).
  1. – = not detected.

Figure 4.

 SEM image of a polished feldspar-rich CAO with feldspar grains embedded in groundmass. Enlarged area is that marked with a black box in the main image. The morphology of the feldspar indicates melting. In the enlarged image, part of a “melted” grain is shown with the components (feldspar-rich interior, contact-rim, mesostasis) marked. The chemical profile (EDX data) shows the diffusion or mixing of the feldspar with the mesostasis material.

Figure 5.

 SEM image of a spherule showing a mixture of both feldspar (FSP) and olivine (OL) grains in the same CAO. The olivine grain (enlarged) shows a dissolution rim. Inner (IR) and outer (OR) rim have different Ca and Al contents, as shown in the diffusion profile.

Figure 6.

 Results of SEM/EDX analyses (in atom %) from the polished samples. The boxes represent 50% of the samples, with the horizontal line inside the box marking the median composition. The vertical lines are the upper and lower ranges for the rest of the grains. The long horizontal lines show the compositions of the starting materials (solid purple line: feldspar, dashed green line: olivine). The enhanced Fe content of the olivine rims is presumed to be from oxidation and dissolution of Fe metal and sulfides. The increase in Mg within the mesostasis and feldspar can only come from dissolution and mixing of olivine; similarly, the enhanced Al in the olivine rims can only come from admixture of feldspar.

The CAO are very porous and friable objects, which made sample preparation difficult: many of the grains were lost during polishing, and so we did not polish the remaining grains to the finish necessary for high-quality EDX analysis. This explains why we only achieved low totals in our analyses, thus requiring normalization of the analyses for inter-comparison of the results. Atom% was selected as the unit for presentation because it gives information in a suitable format to record both chemical and stoichiometric trends.


Looking at the material on the collector pad, it is clear that some of the grains had been melted and formed spherules in the plasma. However, it is also apparent that the whole process was not very efficient, with abundant unmelted relict particles of the starting material remaining. Spherules make up only a minute fraction of the run products, less than 0.1 volume %, based on optical microscopy. We cannot estimate the amount of starting material that was processed into CAO, because a fair proportion of the powder did not pass through the plasma arc. The SEM images of the whole grains reveal spherules with a size range from a few microns to up to approximately 100 μm. Given the difficulties with separating the melt droplets by hand, it was not possible to pick a suite of grains suitable for a size distribution curve. The range overlaps with sizes typical for chondrules (approximately 0.1–1 mm; Brearley and Jones 1998), and is similar to the maximum grain size of the starting material. Exterior views of the CAO demonstrated that reaction between minerals had occurred, with different phases visible on the surface (Figs. 3a–f). In some cases, grains within the CAO have anhedral shapes and are part of a completely melted droplet (Fig. 3a), while in other cases, relict grains remain—usually with rounded features indicating some melting or resorption (Fig. 3b). Abundant Fe-rich dendrites are present in areas of mesostasis (Figs. 3b–e). A possible compound chondrule, where two spheres are melted together, is visible in Fig. 3d. In contrast to the results of Güttler et al. (2008), we did not find any sintered residues among the reaction products. In addition, there was no evidence, from imagery of the final material, that monomineralic CAO formed. The CAO are not simply clumps of grains that have stuck together as they become plastic—they have clearly experienced melting to different extents. Each of the spherules contains a phase that was not one of the original starting materials. Relict feldspars and olivines are surrounded by and embedded within this component, which we liken to the mesostasis or groundmass within chondrules.

SEM images of the spherules after polishing (Figs. 4 and 5) revealed a range of different internal textures, from complete melting to almost unmelted. The selected grains in Figs. 4 and 5 are representative, all grains include at least some mesostasis, and no grain entirely consisting of mesostasis was found. Also, the grains from the five different experimental runs show no obvious difference in their composition and structure. Many of the run products were mixtures of olivine and feldspar (Fig. 5). The CAO in Fig. 4 has abundant bubbles and anhedral (or amorphous) feldspar grains set in groundmass. The boundary between feldspar and mesostasis is diffuse, rather than sharp, suggesting that the feldspars are relict grains that are being resorbed into the mesostasis, rather than crystallizing from it (Fig. 4). Diffuse zones around olivine grains are also common (Fig. 5), signifying that, like the feldspars, the grains show signs of resorption.

Comparison of the chemistry of the run products with that of the starting materials provides further evidence for melting and diffusion (Figs. 6 and 7; Tables 1 and 2). The groundmass that surrounds the feldspars and olivines is a phase with a mixed composition, containing both magnesium and aluminum, implying that it has formed from olivine plus feldspar (labradorite plus forsterite). Mesostasis shows the highest average Fe content of all components (12.9 at. %); while the starting phases contain 2.0 atom% (forsterite) or were iron-free (labradorite). Given the steep increase of Fe in the mesostasis, it is probable that at least some of this Fe is from oxidation of the Fe metal and sulfide in the starting materials. Si abundance remains constant in the feldspar and olivine grains (between 14.5 atom% and 15.3 atom%). Sulfur, manganese, potassium, and nickel contents are generally very low in the starting materials, near detection limits, although K content is enhanced in the mesostasis.

Figure 7.

 Compositional trends within the run products (normalized to 8 oxygen atoms). Si remains constant in the olivine grains (see Fig. 6); Ca, Na, and K are lost from the feldspar and gained by the rims and layers around olivine. The larger feldspar grains in the CAO show signs of dissolution and equilibration with the mesostasis. Ca, Na, and K variations are smaller in the olivines than in the feldspars, with even the outer rims having similar compositions to the starting material. The final mesostasis falls between the end members, but closer to the olivine starting material, indicating a significant olivine component in the mesostasis.

The major element composition of the feldspars does not change much from the starting materials (Fig. 6; Table 2), reinforcing the conclusion from morphology that the grains are relicts, rather than melt products. The anhedral feldspars did not equilibrate with the mesostasis, as shown by the profile from feldspar into the mesostasis (Fig. 4), again indicating partial resorption of the original grains. However, there are some changes in composition: the feldspar grains were entirely Fe-free at the start (labradorite; Table 2), and now contain an average of 1.1 atom% Fe. Sodium content still falls within the range of the sodium in feldspar, but has decreased to 3.4 atom% from 5.1 atom%.

Similar observations of partial dissolution can be made for the olivines (Table 1): the interiors of the grains are approximately uniform in Mg, Fe, and Si, with compositions unchanged from the starting materials. There are increases in Fe, Ca, and Al at a scale of 1–2 μm away from the grain and into the surrounding mesostasis, rims, and an accompanying decrease in Mg (Fig. 5; Table 1). The olivine grain in Fig. 5 is surrounded by narrow inner zone (<1 μm) containing calcium, presumably originating from the mesostasis, although the forsterite in this layer still has good olivine stoichiometry, (Ca0.01 Mg1.6Fe0.3)Si1.06O4. A wider outer zone (approximately 1.5 μm) is a mixture of calcium and aluminum (from feldspar) with magnesium (from olivine). The zones are regions where olivine and feldspar have melted, mixed and then rapidly quenched, preserving the profile.


Our experiments demonstrate that a plasma arc is a viable technique for producing chondrule-like objects from mineral precursor materials. Melted run products are spherical with crystalline mineral grains embedded in a groundmass, implying that gas–grain interaction (heat exchange and collision) in the nebula is a feasible chondrule production mechanism.

However, there are substantial differences between the CAO produced here and natural chondrules. Natural chondrules have several components: silicate grains that have crystallized from a molten state, silicates that have remained unmelted (relict grains), and mesostasis. Iron-nickel metal and metal sulfides are frequently present. The occurrence (or absence) of each of these components is related to the conditions experienced during the chondrule formation event. Thus, the presence of sulfides indicates that temperatures were not held over approximately 1600 °C for longer than a few minutes (Cohen et al. 2004). Metal is an indicator that the conditions were reducing rather than oxidizing. The main silicate components (usually olivine and/or pyroxene) are inferred to have been produced by rapid cooling (up to 3000 °C hr−1 for barred chondrules; Desch and Connolly 2002) from a silicate melt; if relict (olivine) grains are present, then heating above the (olivine) liquidus cannot have been prolonged for more than a few tens of seconds (Connolly et al. 2006).

We can compare features of the CAO produced in the plasma arc with those of natural chondrules. First of all, the spherical appearance of the run products shows that they were mostly solid (at least on the outside) when they were collected—otherwise they would be flattened by impact on the collector. Given the short path length between plasma arc and collector (approximately 10 cm), the run products have cooled at a rate of approximately 107 °C hr−1, orders of magnitude above the upper end of the typical range of cooling rates for chondrules of approximately 3000 °C hr−1 (Desch and Connolly 2002). The rate is also much higher than the temperature regime for the passage of a supersonic shock wave through the nebula, modeled by Morris and Desch (2010). In that model, dust is heated to temperatures above the liquidus, and then cooled very rapidly (>5000 °C hr−1) to subliquidus temperatures, which preserves sulfur and other volatiles. In our experiments, neither metal nor sulfides were preserved: although both iron sulfide and iron metal were present in the starting material, they were absent from the run products, despite the rapid cooling rate. This is not surprising—the experiments were carried out in air, and it is likely that both sulfide and metal were oxidized on heating in the plasma.

In natural chondrules, most of the crystalline phases formed by solidification from a melt. The crystalline grains present in the CAO are a combination of angular and rounded, anhedral grains of both olivine and feldspar. The interiors of most forsterite crystals have essentially the same major element composition as the starting material (Table 1; Fig. 5), implying that they are unchanged, and have simply become incorporated in CAO without melting. They can be regarded as analogous to relict olivine grains in natural chondrules (Jones 1996). There are some signs of partially dissolved forsterite (Fig. 5), but we have so far found no signs of cases where complete olivine grains clearly crystallized from the melt. The kinetics of crystal growth also suggest that formation of olivine crystals in the short time in which the material is cooling from the melt (<0.15 s) would be extremely difficult (e.g., Brandeis and Jaupart 1987), supporting the conclusion that the olivine grains are relict.

A probable explanation for the lack of complete melting of the starting silicates, followed by (re)crystallization, is probably the short time in the plasma experienced by the material.

In the rapid, flash-heating type procedure used in our experiments, congruent melting of the starting materials can be expected. Here, the short heating event produces in a “catastrophic” transformation a melt identical in composition to the starting material. In incongruent melting, a liquid of different composition forms between liquidus and solidus. The kinetics of this process depend on solid-state diffusion, which slows the process down (Greenwood and Hess 1996).

In congruent melting, the growth rate for melt is limited by the kinetics at the solid/melt interface, as silicates produce melts with a high viscosity. Still, grains <100 μm in size completely melt in less than 10−4 s at approximately 1900 °C (Greenwood and Hess 1996); we calculated that material was maintained inside the plasma for approximately 20 × 10−3 s, which should have been sufficient time for melting to occur. The plasma temperature was measured as 2700 °C, implying that despite the extended duration of heating, melting was initiated, but not completed. This inference is supported by the occurrence of very small olivine grains (down to approximately 5 μm) within the CAO—they are not the result of substantial melting of much larger grains because the olivines have the composition of the starting materials, and are surrounded by sub-micron dissolution rims (Fig. 5), indicating that the original grains were small in size.

The diffuse interface between the mesostasis and relict crystalline grains in the CAO is an important difference between the run products and natural chondrules: in natural chondrules, the boundary is usually sharp and well delineated, as a result of rapid crystallization of grains from a melt. Here, the diffuse interface is a reflection of melting and dissolution, not crystallization. Most of the run products are mixtures of different starting materials, as shown by the occurrence of olivine and feldspar grains in the same droplet (e.g., Fig. 5). This points toward collisions between particles, followed by mixing and melting in the very short time interval the grains remain inside the plasma. Particle collisions require convection, which not only brings particles together, but also extends the time the material spends inside the plasma arc, thus the calculated heating times are a minimum. This is an important observation, because it implies that chondrules could form from collisions between separate particles, rather than requiring “dust ball” or dust aggregates as precursors (Jones et al. 2005).

As the element profiles and chemical composition show (Figs. 5 and 6; Table 1), some of the olivine grains are surrounded by a transition zone to the mesostasis with slightly enhanced abundances of sodium, aluminum, and calcium. Slight Ca-enrichment in the inner layer of olivine is widespread, but with the basic stoichiometry still similar to olivine.

Increasing Al in the outer layer marks a smooth transition from the inner zone into the mesostasis. The layering of the dissolution zones follows increasing enrichment in Fe, which is frequently observed in relict grains in chondrules, where Fe-enriched rims surround forsteritic cores (Jones 1996). One interpretation of the profile is that the differences in element contents result from diffusion of Fe outward from olivine into the mesostasis and of Na, Al, and Ca inward from the mesostasis into the layers in direct contact to the olivine grains, but not the actual crystalline olivine (Table 1). Calculation of Ca diffusion into forsterite at high temperatures (1400–1500 °C) and in the given time frame for the melted/heated grain of <<1 s results in diffusion distances of Ca into the olivine grains at least two orders of magnitude below the observed range of 1–2 μm (Morioka 1981; Coogan et al. 2005). Thus, the variations in composition shown so distinctly in Fig. 6 are brought about by mixing between partially melted feldspar/mesostasis and olivine grains (Fig. 7). This is a different process from that in natural chondrules, where solid-state diffusion is thought to be the mechanism by which elements equilibrate.

Most of the feldspar contains low, but significant, abundances of iron and magnesium: the starting composition changed from (Ca,Na)0.9(Al,Si)4O8, to a median composition of (Fe,Ca,Na)0.8(Al,Si)4O8 in the run products. The additional iron probably came from the iron metal and sulfides as well as from olivine: the iron metal powder had a grain size of <1 μm and so oxidized or melted more readily than the silicates, leading to the elevated and variable iron contents of the mesostasis. Figure 7 shows how the major element contents of the olivine and feldspar change from the starting materials to the run products, producing an intergrain mesostasis with the stoichiometry and composition of a mixture between olivine and feldspar. Again, this can only have been produced by melting of the materials together. The loss of volatile Na (Figs. 6 and 7), coupled with their anhedral to amorphous structures, shows that most of the feldspar grains partially melted. The implication is that the temperature of the plasma at least reached the eutectic temperature of an anorthite-forsterite mix (approximately 1266 °C; Longhi 1987). Melting is also indicated by the occurrence of rounded voids in the CAO, both within mesostasis and within silicate crystals. Such voids are not common in natural chondrules, and are signs of gas bubbles, possibly sulfur dioxide, and maybe other volatile species such as sodium from the feldspar. The presence of these voids shows that (partial) melting must have taken place, as they could not form in a solid matrix.

The observed features in the experiments also show similarity to such in cosmic spherules and micrometeorites. Especially cosmic spherules (CSs) and scoriaceous micrometeorites (SCMMs) are of interest, as they melted entirely (CS) or partially (SCMM) during atmospheric entry (Genge et al. 2008).

Iron-rich dendrites (also called magnetite/spinel rims) as observed in Figs. 3b–d occur in G-type spherules, micrometeorites with iron-rich dendrites in a silica-rich matrix. Extreme cases, like spherules dominated by such dendrites are I-type spherules, which are very similar in structure to the grain in Fig. 3e (Taylor et al. 2000; Genge et al. 2008; Van Ginneken et al. 2010).

Relict olivine grains in a (probably) iron-enriched groundmass formed from melted materials are observed in the coarse-grained S-type micrometeorites (Taylor et al. 2000; Genge et al. 2008).

SCMMs exhibit bubbles or voids in various sizes, which can also be found in the other CS-type micrometeorites. In particular, vesicular, relict-bearing SCMMs with olivines in void-rich mesostasis and interstitial glass are very similar to the results from this study. Also observed in SCMMs are rims of Fe-rich pyroxenes around relict Mg-rich pyroxenes. These could be similar to the Fe-enriched mixing zones around olivines observed in this study (Genge et al. 2008).

The very low sulfur contents in the mesostasis (Table 1), as well as the lack of any relict sulfides, may also explain the formation of the vesicles, as also observed for SCMMs by Taylor et al. (2011). Here, the decomposition of sulfides is identified as the source of gases trapped in the melt.

Experiments to simulate the atmospheric re-entry of micrometeorites were made using flash heating of Orgueil and Murchison particles 200–400 μm in size by heated air/CO2/N2 gas in a furnaces, which were quenched in air (Toppani et al. 2001; Toppani and Libourel 2003). For temperatures from 1200 to 1500 °C, with duration of 2–120 s, the run products feature iron-rich spinel in a glassy matrix similar to this study. Our findings possibly allow the formation of this feature at shorter heating times (approximately 0.02 s), but with higher temperatures (approximately 2700 °C), which confirms the model presented in Toppani et al. (2001), which requires temperatures of over approximately 1500 °C to form CS or SCMM at very short times <<1 s.

Conclusions and Next Steps

The aim of this project was to test the hypothesis that accelerating material through a fixed plasma field might be a valid simulation of a supersonic shock wave traveling through a cloud of gas and dust. We were able to produce spherical objects with generic similarity to natural chondrules—crystalline silicates embedded in a feldspar-rich “mesostasis,” in the size range of up to 100 μm. Further features observed in the experimental run products (zoning, rounded outlines, feldspar phenocrysts) are not directly analogous to those found in natural chondrules. The chemical composition (especially Fe content) of feldspars, as well as voids from gases, indicates that melting of starting feldspar was common. All observed olivines are regarded as relict grains, with only their rims showing signs of melting/mixing with the surrounding mesostasis, as indicated by changes in Al, Ca, and Na contents.

Some differences in reaction products from natural chondrules were to be expected, as the experimental conditions did not directly replicate those under which chondrules were produced. It should be kept in mind that the study was designed as a “proof-of-principle,” so although the selected starting materials (feldspar, olivine, metal, and sulfides), and their grain size (<100 μm) were reasonably close to natural chondrules, several experimental parameters were orders of magnitude away from the presumed environment of the early solar system. This is especially true regarding gas composition and pressure—the extremely oxidizing atmosphere of the experiment chamber (oxygen-rich atmosphere at ambient pressure) is very different from the reducing hydrogen-rich solar nebula with a pressure of approximately 10−4 to 10−6 bar. The heating time of the material (approximately 0.02 s) is very short compared with the >>10 s observed for natural chondrules (Connolly et al. 2006), even though mixing between heated and melted starting materials indicates convection within the plasma arc, which in turn would have extended the residence time of material in the heated region. The cooling rate in the experiments (approximately 107 °C hr−1) proved to be much higher than the values calculated in models for peak cooling rates in shock heating (approximately 104 °C hr−1; Morris and Desch 2010) or based on chondrules (approximately 103 °C hr−1; Desch and Connolly 2002). We infer, from the textures of the run products, that the peak temperature (approximately 2700 °C) was clearly above that required for complete melting to occur, and it was this factor that had the greatest influence over the final run products.

Our results essentially produced chondrule-like objects through the bombardment of dust grains by ionized gas molecules and plasma, a scenario that we have likened to the passage of a shock wave through nebula gas. However, the process is also akin to that which would occur during lightning strikes, and so our results do not rule out chondrule formation by nebular lightning (Horányi et al. 1995; Desch and Cuzzi 2000; Güttler et al. 2008). The production of CAO with mixed mineralogy from monomineralic starting materials also shows that collisions between particles are an important mechanism within the chondrule formation process, such that dust aggregates are not necessarily required as chondrule precursors.

The experiments are suitable as analog process for the atmospheric entry of micrometeorites. Here, features like vesicles or iron-rich dendrites similar to partially or completely melted micrometeorites were reproduced. This confirms earlier models (Toppani et al. 2001) and helps to define the formation condition at high temperatures and very short heating.

Despite the limitations of the experiment, there are now clear directions for us to pursue the work. The first parameters to change in future syntheses are the environmental conditions (pressure, gas composition, etc.), which will be adjusted to more realistic values appropriate for the solar nebula. Thus, the next round of experiments will be conducted in an environment chamber at low pressure, such that the environment is reducing and not oxidizing. We also intend to vary our starting material to be more realistic, using ground natural chondrules. We plan to develop confined plasma, so that material has a larger path-length through the plasma, allowing an extended time for reaction. In the experiments described here, we assumed that plasma was a valid analog for the nebula gas. However, we recognize that there are sufficient differences between plasma and a gas that the analogy might not hold, thus one of our future experiments will replace the plasma with heated gas. We also plan to gain a more exact control of the heating time of the material by using acoustic waves to levitate dust grains within the gas.

Acknowledgments— Thanks to Diane Johnson for assistance with the SEM. Many thanks also to Roger Hewins (Paris) for helping improve the original manuscript substantially, and to Rhian Jones (University of New Mexico), Masayuki Uesugi (Osaka), and Guy Libourel (Nancy) for their reviewers’ comments. Financial support from the STFC is gratefully acknowledged (Grant No. ST/F003102/1). This is a publication from CEPSAR, the Centre for Earth, Planetary, Space, and Astronomy Research at the Open University.

Editorial Handling— Dr. Edward Scott