Abstract– Tektites, natural silica-rich glasses produced during impact events, commonly contain bubbles. The paper reviews published data on pressure and composition of a gas phase contained in the tektite bubbles and data on other volatile compounds which can be released from tektites by either high-temperature melting or by crushing or milling under vacuum. Gas extraction from tektites using high-temperature melting generally produced higher gas yield and different gas composition than the low-temperature extraction using crushing or milling under vacuum. The high-temperature extraction obviously releases volatiles not only from the bubbles, but also volatile compounds contained directly in the glass. Moreover, the gas composition can be modified by reactions between the released gases and the glass melt. Published data indicate that besides CO2 and/or CO in the bubbles, another carbon reservoir is present directly in the tektite glass. To clarify the problem of carbon content and carbon isotopic composition of the tektite glass, three samples from the Central European tektite strewn field—moldavites—were analyzed. The samples contained only 35–41 ppm C with δ13C values in the range from −28.5 to −29.9‰ VPDB. This indicates that terrestrial organic matter was a dominant carbon source during moldavite formation.
Tektites are natural silica-rich glasses produced by impact events. Individual tektites attain typically the size of max. several centimeters across; however, both considerably smaller and substantially larger samples have been found. Tektites are usually subdivided into splash-form and layered types (the terms layered tektites, Muong Nong tektites, or Muong Nong-type tektites are considered as equivalents in this paper). On the Earth, they occur in geographically confined areas called strewn fields. There are four principal strewn fields currently defined (Koeberl  and references therein): the North American tektite strewn field, 35.5 Ma in age (associated with the Chesapeake Bay impact structure in the United States); the Central European tektite (CET) strewn field, 14.7 Ma in age (associated with the Ries crater in Germany); the Ivory Coast tektite strewn field, 1.1 Ma in age (associated with the Bosumtwi crater in Ghana); and the 0.78 Ma Australasian tektite strewn field for which no source crater has been identified yet. Tektites (both splash-form and layered) from the CET strewn field are traditionally called moldavites, which is a term used also throughout this article.
One typical feature of tektite glass is the occurrence of bubbles. These bubbles are usually hundredths of millimeter to several millimeters in diameter, rarely exceeding one centimeter. Their shape is either spherical or lenticular. Some bubbles form flat shapes elongated in the direction of glass flow (Bouška 1994). In the tektite glass they are distributed either randomly or in clusters, sometimes arranged in rows. Small bubbles in moldavites frequently concentrate around lechatelierite grains (silica glass particles), which led some authors (e.g., Dolgov et al. 1969) to distinguish between large bubbles randomly distributed in glass, and smaller bubbles occurring in lechatelierite or around it. The main purpose of this study was to review the published literature on the composition and pressure of gases contained in the tektite bubbles with the main focus on carbonaceous compounds (noble gases are not covered).
After the first attempt to determine gas pressure and composition in these bubbles (Beck 1910), many studies employed a wide range of extraction and analytical methods to obtain the bulk composition of gas phase contained in tektites (see Table 1). The present authors tried to collect all available published information on gas composition in tektites, including papers contained in local journals and written in languages other than English.
Table 1. A review of published data on gas content and gas composition in tektites. Papers specifically devoted to noble gases were not included.
Data source (reference)
Sample type and size, method of gas extraction
Estimated pressure in bubbles
Gas phase composition (vol%)
The pressures in bubbles reported in individual cited papers are given in the original units. For approximate conversion:
Various tektites, contaminated surface layer strongly removed by abrasion; platinum crucibles cleaned by heating under oxidizing conditions; outgassing 100–200 °C, induction-heating of samples up to 1450 °C; noncondensable gas volume measurement; water converted to hydrogen and measured both hydrogen yield and its isotopic composition
The water yield obtained by tektite melting was below 0.014 wt%, (the whole data range from 0.014 to 0.0003 wt% H2O, the lowest in moldavite)
Only H2O was obtained; the quantity of noncondensable gases was below 0.01 cm3 STP in all samples; noncondensable gases not further analyzed
Author concluded that H2, CO a CO2 given in the older papers were formed by a reaction between water vapor and the traces of carbon on crucibles during extraction; the line of Friedman (1958) was not heated as a whole, a proportion of water could have been derived from the line internal surfaces
8 samples (5 moldavites, 1 philippinite, 1 indochinite, 1 australite); cut plates 0.5–1 mm thick containing bubbles; crushing within a liquid at normal pressure, then micromanipulation of bubbles and volumetric absorption microchemical method
For the moldavites the volume of bubble decreased after crushing 610–2980 times
In moldavites found 53–69% CO2; 22–40% H2; 2.6–4.8% of the compounds from the group H2S + SO2 + NH3 + HCl + HF; 0.2–6.1% of N2 and noble gases; CO not detected; in other tektites also dominance of CO2, H2 up to 8.8%, no N2
Tektites of the Muong Nong type from Indochina and Libyan Desert glass, outgassing 12 h(?) at 120 °C under vacuum; gradual crushing under vacuum at laboratory temperature and online sensitive mass spectrometry (quadrupole)
Bubble pressures in order of 100 mmHg (13,332 Pa)
Bubbles in Muong Nong tektites dominated by CO2, concentrations of O2, CO2, CO, and SO2 were strongly different even between adjacent bubbles; the CO2/CO ratio was about 25; ratios N2:Ar:Kr:Xe were atmospheric
Irghizites and impact glasses from the Zhamanshin crater; both thermal extraction at 1100–1500 °C and crushing under vacuum at 20 °C; followed by mass spectrometry
At the thermal extraction sample weight loss was 0.09% for the acid irghizite type, and 1.67% for the basic irghizite type
During crushing the acid irghizite types produced 20–40 times more gas than the basic irghizite types
Thermal extraction, acid types: H2O, CO2, N2, at a higher temperature O2; basic types: H2O, CO, CO2, SO2, at a higher temperature O2; extraction by crushing (both types): N2, CO2, probably also CH4 and HCl
A 5 cm3 large bubble in philippinite, crushing under vacuum, probably at usual laboratory temperature
Pressure in the bubble estimated at 10−4 atm (10 Pa)
Released gases dominated by H2 (approximately 90%), further detected H2O, in traces CH4 and noble gases; N2, O2, CO2 not detected
With respect to gaseous carbon compounds in tektites, the published data revealed the presence of CO, CO2, CH4, and other hydrocarbons. Analysis of the published data contained in Table 1 indicates that the CO/CO2 ratio and also the gas yield during extraction does, at least partly, depend on the method of tektite cleaning before the analysis, and on the method of gas extraction (usually either high-temperature vacuum-melting, or low-temperature vacuum-crushing). This indicates that besides the gaseous carbon species contained in the gas phase in bubbles, tektites also contain other portions of carbon contained in the glass, either in the form of a solid phase or in the form of physically dissolved gas. These carbon compounds can be released during sample melting but cannot be volatilized during sample crushing at a low temperature.
The conclusions obtained from the review of the published literature directly initiated the analytical part of this study. Determinations of carbon contents and carbon isotopic compositions of the tektite glass were the next logical steps. These data are necessary to interpret differences in the composition of gases obtained by tektite melting and by tektite crushing methods. Moldavites—tektites from the CET strewn field—were selected for this pilot analytical study. The study was supplemented by analyses of organic carbon and carbonate carbon contents in the surface Neogene sediments of the Ries target area.
A Review of Available Data on Volatile Compound Content and Composition in Tektites
Mechanisms of Formation of Bubbles
The number of bubbles is lower in tektites with high temperature of formation (Barnes 1964, 1969; Konta and Mráz 1969; Trnka and Houzar 2002). Bubbles in the tektite types with high temperature of formation are usually more spherical. Most authors considered the formation of bubbles to be a result of internal gas pressure during tektite cooling in an environment where the external pressure was very low (e.g., Suess 1951; Chao 1963). In fact, several mechanisms of bubble formation may have been involved. As a result of this, even a single tektite sample may contain several populations of bubbles formed by different mechanisms and containing gases of different compositions and pressures (Jessberger and Gentner 1972).
The first, probably rather rare mechanism of bubble formation is mechanical entrapment of (very low-pressure) external atmosphere in the area where the tektite glass solidified. During experiments with glass similar to tektites, Hawkins (1963) found that small bubbles of this type display characteristic size distribution with a steep gradient in the size versus frequency plot.
The second bubble type is represented by bubbles formed by gases which were physically dissolved in the glass and released to form a gaseous phase during glass cooling above the temperature of transformation, or during a decrease in external pressure. These bubbles show a less steep gradient in the size versus frequency plot (Hawkins 1963) and display a size limit below which they cannot be formed because of surface tension of the molten glass.
The third possibility of bubble formation results from thermal volumetric changes of glass during cooling. When the surface of a tektite is already below the transformation temperature and its internal part remains still plastic, thermal volumetric contraction of the internal part can be compensated by a bubble formation. This is given by the fact that thermal volumetric expansion of molten glass is higher than that of glass below the temperature of transformation (Bouška et al. 1993). The internal pressure in bubbles of this type can be (if the glass does not release any gas) almost a vacuum. Pressures in bubbles of this third type have no relationship to external pressure during the tektite glass transformation from molten to solid. Formation of bubbles with extremely low internal pressure is well known also from the technical practice, during glass production, when molten glass is rapidly cooled. With respect to the very small thermal volumetric expansion of tektite glass, bubbles of this type should be very small. The coefficient of thermal volumetric expansion of the studied moldavite glass is small (O’Keefe 1976), similar to that of artificial borosilicate glass of the Pyrex type. Moldavite glass is therefore much more resistant to thermal shocks than usual commercial alkali glasses.
Internal Pressure of Gases Contained in the Bubbles
Gas pressures in the tektite bubbles reported in most papers are significantly lower than the standard atmospheric pressure. In some bubbles, virtually no gas was found. Martin (1934) reported gas pressures in tektite bubbles below 5 mmHg (approximately 666 Pa). Suess (1951) found pressures of only thousandths of atmosphere (hundreds of Pa) in bubbles in a splash-form tektite from the Philippines. Zähringer (1963) studied bubbles of eight splash-form tektites from the Philippines and reported pressures of thousandths, hundredths, and also tenths mmHg, reaching 1.9 mmHg (approximately 253 Pa) in one case. The same author measured a pressure of 40 mmHg (approximately 5333 Pa) in a layered tektite type from Laos. Rost (1964, 1972) measured a pressure of 33 mmHg (approximately 4400 Pa) in a large bubble in moldavite. Koeberl (1988) stated that bubbles in the layered tektites contained gases at much higher pressures (up to one-third of the atmospheric pressure; approximately 33,800 Pa) than the splash-form tektites. Matsuda et al. (1995, 1996) estimated the pressure in a large bubble in a splash-form tektite from the Philippines to be about 10−4 atm (approximately 10 Pa) based on the gas yield (mostly hydrogen).
With respect to several possible mechanisms of bubble formation (see the Mechanisms of Formation of Bubbles section) any estimates on the external atmospheric pressure during tektite solidification based on pressure in bubbles are highly problematic. It should be noted that during the tektite cooling below the temperature of glass transformation, the gas pressure decrease in a bubble, resulting from the equation of state, drops to about one-fifth of the original pressure. The temperature of melting of studied moldavite glass is above 1100 °C according to Bouška and Konta (1990). Heide and Brückner (1971) reported the temperature of transformation of moldavite glass between 900 and 1030 °C.
Some bubbles in tektite can communicate with surface via narrow channels or cracks. Great care must therefore be taken during the sample outgassing prior to melting/crushing to pump off all gases from these cavities connected with the tektite surface.
Composition of Gases Released from Tektites
Data on the composition of volatile compounds (gases) released from tektites by several different methods can be found in a number of papers starting from Beck (1910). The variation in the published gas compositions (and gas yields) is wide. Gas composition similar to the normal Earth’s atmosphere with dominance of nitrogen and presence of oxygen was found quite sporadically (e.g., by Müller and Gentner 1968). Oxygen was found also by O’Keefe et al. (1962). All other published articles (see Table 1) reported gas composition strikingly different from the Earth’s atmosphere, with nitrogen and oxygen either totally absent, or present in only very small amounts. The absence of oxygen is predictable with respect to the reduced character of glass, with a dominance of ferrous iron over ferric iron (Dunlap et al. 1998).
Barnes and Russel (1966) observed bubble collapses and glass devitrification around them in tektites which were heated for 3 days by accidental fire of a house in Santa Mesa (Manila, Philippines), in which they were stored. They concluded that bubbles were originally formed by water vapor during the tektite formation, and that this increased water content around and in the bubbles facilitated glass devitrification during the fire. This process with a similar bubble behavior was simulated by laboratory heating of tektites from the Philippines for 4 days at 825 °C. On the other hand, neither collapse nor devitrification was observed around bubbles in laboratory-heated layered tektites (Barnes and Russel 1966). Among other authors, only Friedman (1958) considered H2O as an important component in the bubbles in tektites. It is generally known that water content in tektite glass is very low (lowest for moldavites), and that water vapor as the main component in bubbles is therefore not probable in such extremely dry glass types. Dolgov et al. (1969b) heated plates cut from moldavites containing bubbles to temperature above 1000 °C and observed no changes of the bubble walls. The expected melting of glass on internal surfaces of bubbles was not observed. During cooling of the samples down to −170 °C, no condensation inside the bubbles was observed.
Most of the experiments reviewed in Table 1 indicated a gas mixture with the dominance of CO, CO2, and H2 in the tektite bubbles. The studies using gas extraction based on tektite melting regularly found a significant presence of CO in the released gases, while the papers using sample crushing at a low temperature either did not detect CO at all, or only in a small amount. In the case of crushing or milling extraction, the main carbonaceous component was usually CO2.
The gas yield is generally higher during the tektite melting experiments than during the low-temperature crushing or milling (see data in Table 1). When thermal extraction was employed, the gas yield was usually around several tenths cm3 of gas (at standard temperature and pressure, STP) per gram of tektite. If the gas is CO + CO2-dominated, this corresponds to several tenths mg of carbon per gram of tektite. Carbon yield could have been elevated by surface tektite contamination in cases when the surface tektite layer was not removed prior to the analysis. The gas yield during crushing was always lower by an order of magnitude. Only several mm3 of gas (STP) was usually obtained from samples several grams in weight.
The difference in the composition of gases released by tektite melting versus tektite crushing cannot be explained by chemical reactions during the extraction procedure (e.g., reduction of CO2 to CO by ferrous iron in the glass), as the total gas yield of both methods differs. Data indicate that, besides carbon contained in the gaseous phase in the bubbles, the tektites contain another carbon reservoir directly in the glass. This is in agreement with the statement of O’Keefe (1976) that tektites contain usually 50–150 ppm carbon.
Some authors detected reduced carbonaceous compounds directly in the tektite glass (Petersile et al. 1967; Dolgov et al. 1969; Muenow et al. 1971). Petersile et al. (1967) extracted reduced carbonaceous compounds from the crushed moldavite glass by usual chloroform extraction at temperature of 53–58 °C for 60 h, with a yield of 0.0072 wt% of organic compounds. Based on the absence of humic acids, Petersile et al. (1967) supposed that the extracted organics are not related to secondary sample contamination, but represent compounds which underwent high-temperature processes. Frank et al. (2003) unsuccessfully searched moldavites and impact glass from the Ries crater for fullerenes.
Some portion of carbon in tektites can be bound to Si-C bonds or to compounds similar to organosilans or organosiloxans. Zbik et al. (2000) detected hydrocarbon fragments on the internal surface of gas bubbles in irghizite, which were similar to those usually detected in organosilans or organosiloxans. In this respect, it is noteworthy that some tektites release a small quantity of oxygen at very high temperatures (approximately 1500 °C) above the glass melting temperature (Heide et al. 1981 for irghizite). Silicon-carbide was observed in impact melts from the Ries crater by Hough et al. (1995).
Another indirect support for the presence of reduced carbonaceous compounds in the tektite glass was a description of the presence of elemental Si in moldavites (Cílek 1985; Cílek and Frich-Char 1988). Elemental Si is technically produced by a reduction of SiO2 using a graphite electrode and electric arc. The presence of elemental Si in moldavites could be a result of high-temperature reduction of SiO2 by carbon.
General Characteristics of Moldavites and Their Strewn Field
Most of the Central European tektites—moldavites—possess a splash-form shape without ablation features on their surface. Layered or brecciated moldavites occur rarely (Glass et al. 1990; Švardalová 2007). Typical appearance of splash-form and layered moldavites is shown in Fig. 1. Most of the known localities are distributed in the Czech Republic with some less frequent finds in adjacent parts of Germany and Austria. Micro-tektites have not been found. Moldavites have been found in several discrete regionally limited areas—substrewn fields. These partial areas in fact represent a combination of the original distribution in the strewn field and the subsequent removal and redistribution of moldavites by fluvial erosion and transport (Fig. 2). In the Czech Republic, moldavites occur in three main areas located in South Bohemia, Moravia, and West Bohemia. Rare finds of moldavites beyond the limits of the major substrewn fields represent either samples transported by surface fluvial processes over a long distance or very small remains of other original substrewn fields (Žebera 1972; Bouška et al. 1999; Trnka and Houzar 2002; Žák 2009).
The largest preserved part of the CET strewn field is located in South Bohemia, in a region that is bounded by the towns of Písek, Veselí nad Lužnicí, České Velenice, Český Krumlov, and Prachatice. The oldest moldavite-bearing sediments (sandy clays and/or clayey sands, gravels, and sands) in the region are usually considered to be of Middle Miocene age. They were formed in different depositional environments along the shores of freshwater lakes, which were sediment-filled at that time (Žebera 1967, 1977). Moldavites were commonly redeposited into Pliocene to recent fluvial sediments, and into Pleistocene solifluction soils (Bouška 1972). Unusual moldavites with extremely high SiO2 contents, strikingly light green color, relatively low content of lechatelierite, and high average weight were found at a few localities in the northern part of the South Bohemian substrewn field near Radomilice (Bouška 1997).
In Western Moravia, moldavites have been reported from the area defined by the towns of Třebíč, Moravské Budějovice, Znojmo, Hrušovany nad Jevišovkou, Ivančice, and Náměšť nad Oslavou. The oldest moldavite-bearing sediments in Moravia are of Middle to Upper Miocene age and correspond to colluvio-fluvial sand-dominated sediments with gravel admixture. Near the town of Znojmo, moldavites have been found in reddish colored, fluvial, coarse-grained sandy gravels of Pliocene age. The youngest moldavite-bearing sediments in Moravia are Quaternary colluvial loams and Pleistocene terrace sandy gravels, developed along the present-day rivers.
Moldavite-bearing sediments of Western Bohemia (Cheb Basin, near the town of Cheb) are of Pliocene age. These are fluvio-lacustrine, poorly sorted gravels and sands (Bouška et al. 1995). Chemical analyses and colors of the Cheb Basin moldavites correspond to those of the moldavites from South Bohemia (Bouška et al. 1995; Řanda et al. 2008; Skála et al. 2009). It is obvious that local moldavite finds constitute a separate substrewn field, the extent of which has not possibly been fully constrained yet.
Moldavites are known also from the Miocene fluvial sandy gravels and from fluvial sediments of the Lower Pleistocene age in Lusatia, northeast of Dresden, Germany (Rost et al. 1979; Lange and Wagner 1992; Lange and Suhr 1999). The character of the Lusatian moldavites excludes fluvial transport from the South Bohemian substrewn field, and together with differences in the chemical composition, it substantiates the existence of a separate substrewn field (Lange 1995, 1996).
In Austria, moldavites occur in the Horn area (Suess 1914; Koeberl et al. 1988). The moldavite-bearing sediments are Miocene fluvial gravels and sands. Local moldavites show a relatively wide compositional range and their colors are similar to those occurring in South Bohemia.
Moldavites possess several specific features among tektites. They are the most silicic tektites with SiO2 content ranging typically between 75 and 85 wt%. Another unusual feature of moldavites is the dominance of K2O over Na2O. Unlike in all other tektites, silica content does not negatively correlate well with each of other major elements (Meisel et al. 1997). The contents of TiO2 and FeO are relatively low, which results in a high translucency of moldavites. It should be also noted that the composition of moldavites is not uniform. Microprobe data reveal a substantial scatter in chemical compositions, not only among individual samples, but also on the micrometer scale within an individual specimen (e.g., Engelhardt et al. 2005; Skála et al. 2009). Recent studies (Dunlap et al. 1998; Rossano et al. 1999; Skála et al. 2009) confirmed that moldavites contain <6% of the total iron in ferric form.
Volatile components are extremely depleted in all tektites. Water content in moldavites varies between 0.006 and 0.010 wt% (Beran and Koeberl 1997). This has recently been confirmed also by confocal Raman spectrometry with measured values about 0.010 wt% (Thomas, personal communication; see also Thomas et al. 2008). Chondrite-normalized rare earth elements (REE) patterns are typical of mature Phanerozoic sediments with light REE considerably enriched over heavy REE (LaN/YbN ∼ 10–14; CeN/YbN ∼ 7–11) and negative europium anomalies (Eu/Eu* typically approximately 0.65–0.70; Řanda et al. 2008).
The size of the individual moldavite samples is not uniform throughout the entire CET strewn field. The largest moldavites generally occur in the Moravian part of the strewn field whereas the smallest are from the Cheb Basin. The heaviest moldavite known, 258.5 g in weight, was found near Slavice in Moravia. Current shapes of moldavites reflect both the primary shapes generated during the moldavite formation and later geological processes. Surface sculpting consisting of pits and grooves was usually formed by low-temperature chemical corrosion during the storage of moldavites in sediments. Postdepositional origin of the sculpting is confirmed by its presence on both original tektite surfaces and on surfaces produced by tektite breakage either during the fall or after the deposition in sediments. In addition to chemical corrosion, moldavite morphology was also distinctly influenced by mechanical abrasion. New sculpting commonly evolved on the surface of mechanically abraded moldavites after their deposition in a new type of sediments. The color of moldavites ranges from pale green to brown. Moldavites, contrary to other tektite groups, are translucent or transparent. Compared with other natural glasses, tektites are in general fairly homogeneous, being almost devoid of unmelted mineral grains and crystallites. Although they are rarely present, crystalline inclusions are formed by microscopic magnetite spherules, sometimes containing a native iron core or an admixture of wüstite. Small hematite particles, irregular fragmented grains of coesite, and baddeleyite are rare (Trnka and Houzar 2002). The above-mentioned small-scale chemical heterogeneity usually correlates with the fluidal internal fabric in the form of schlieren; this fabric is further accentuated by small bubbles and lechatelierite (silica glass) particles arranged parallel to the overall texture.
The bubbles are omnipresent in moldavites (Fig. 3). They typically constitute about 0.1% of the volume of individual splash-form moldavites (Trnka and Houzar 2002). With respect to differences in abundance, size, and shape of bubbles in individual substrewn fields, Barnes (1964) stated—based on a relatively small number of samples—that Moravian moldavites, as compared with Bohemian moldavites, contain a lower number of bubbles, which are smaller and more spherical. Similarly Lange (1995), based on the study of a relatively small number of samples, also concluded that the relative frequency of bubbles decreases from Bohemian to Lusatian and to Moravian moldavites. Based on a review of all available data and a study of a large number of specimens, Trnka and Houzar (2002) confirmed that Moravian moldavites and moldavites from the Radomilice area in southern Bohemia have fewer bubbles than typical South Bohemian moldavites. The bubbles in layered moldavites (which are generally rare in the CET strewn field) are much more frequent (Trnka 2003). Švardalová (2007) reported bubble contents of max. 4 vol% for the layered moldavites.
Typical bubble dimensions in moldavites range from a hundredth to tenths of mm; closed bubbles exceeding 1 cm are very rare. Nevertheless, the surface of some moldavites reveals features that may represent former, currently open bubbles, which attain a few centimeters across. The shape of the bubbles in the matrix is either spherical or elongated and flattened at the same time. A variable amount of small spherical bubbles occurs in lechatelierite. In the layered moldavites, the density of these bubbles can result in a frothy appearance of the lechatelierite particles. Such foamy lechatelierite was found in some South Bohemian moldavites, whereas none was observed in the Moravian moldavites.
Absolute datings of moldavites and the Ries crater have most recently been extended by Di Vincenzo and Skála (2009) and Buchner et al. (2010), respectively. The former authors obtained a mean Ar-Ar age of 14.68 ± 0.11 Ma (2σ) for 7 moldavites sampled from the territory of the Czech Republic. The latter ones suggested an age of 14.59 ± 0.20 Ma (2σ) as the best value for the Ries impact event. Independent reliable stratigraphic constraints on moldavite age are missing. Nevertheless, the age of the Ries crater (and thus moldavites) was determined magnetostratigraphically (Aziz et al. 2008) to lie within the 14.58–14.78 Ma age range (polarity chron C5ADr; polarity chron is a fundamental unit of geologic time, magnetic polarity-chronologic unit).
Numerous geochemical studies excluded both the Hercynian crystalline basement and Mesozoic and Tertiary marine sediments as a potential source material of the moldavite formation. Currently, the Miocene Upper Freshwater Molasse sediments (“Obere Süßwasser Molasse” in German; OSM) are considered to have contributed substantially to moldavite melt (e.g., Bouška et al. 1973; Luft 1983; Horn et al. 1985; Engelhardt et al. 1987, 2005; Meisel et al. 1997; Trnka and Houzar 2002; and references therein). Although the overall geochemical characteristics point to the origin from the OSM sediments, there is no direct counterpart to moldavites among OSM sediments sampled in or around the Ries crater. The chemistry of moldavites, like all tektites and impact glasses, was therefore interpreted as a result of melting and incomplete mixing of several components (Delano and Lindsley 1982; Delano et al. 1988; Meisel et al. 1997; Magna et al. 2011).
Recently, Řanda et al. (2008) argued that, in addition to the traditionally accepted source materials (silica-rich sands, marls/clays, and Ca-/Mg-carbonates), burned organic matter (trees, shrubs, and soils) should be considered as the fourth component of the moldavite source material. This hypothesis was based mainly on the extremely high K/Na ratio, and significant enrichment in Ca, Mg, Mn, and their positive correlations with K, which were observed in some moldavites from the Cheb Basin. Independently, similar results (not their interpretation) have been presented by Skála et al. (2009) in a statistical analysis of geochemical data from a different set of moldavite samples from the Cheb Basin. The existence of a biogenic component in source materials for tektites was assumed earlier by Kinnunen (1990) who suggested that siliceous phytoliths—biogenic opal formed in plant tissues—was the precursor of lechatelierite grains in tektites.
Whatever the actual precursors were, their lithological variations, facies changes, as well as lateral and vertical distribution in the target area are reflected in the variation of the observed moldavite properties over the entire moldavite strewn field. The strewn field properties also reflect the dynamics of moldavite transport from the Ries area to the strewn field.
Three typical moldavite samples were selected for this study: one from the Moravian substrewn field (No. 151, Moravské Budějovice locality), another from the northwestern part of the South Bohemian substrewn field (No. 152, Radošovice locality), and the third from the southeastern part of the South Bohemian substrewn filed (No. 153, Slavče locality). General characteristics of the samples are contained in Tables 2 and 3; positions of the sampled localities are shown in Fig. 2.
Table 2. Chemical composition of the studied moldavites, determined by INAA.
Data are given in wt% for major oxides and in ppm for other elements; the element ratios are dimensionless. Because of the relatively big uncertainty of Si determinations by INAA (up to 2% rel.), data on Si as the main component of moldavites were calculated as a difference between the sum of other major oxides and the total of 100 wt%.
Table 3. Weight characteristics of the studied moldavites, weight loss after cleaning, carbon content, and carbon isotopic composition of moldavite glass.
Weight after cleaning before etching (g)
Weight after etching; 4 h 5% HF (g)
Weight loss during etching (rel.%)
C content; C/S Analyzer Eltra (ppm)
C content; FC-GC-IRMS on line (ppm)
δ13C; FC-GC-IRMS (‰ VPDB)
Sample load preparation line (mg)
C content; calculation, Pirani gauge preparation line and MS signal (ppm)
δ13C; preparation line (‰ VPDB)
Czech Republic, Moravia, Moravské Budějovice
Flat shard with lustrous fracture surfaces, party conchoidal, 25.5 × 18.0 × 10.0 mm
approximately 30 ± 10
Approximately −26 ± 2 (low signal)
Czech Republic, Southern Bohemia, Radošovice
Flat pebble with shallow, finely sculpted holes, light green, 19.5 × 15.0 × 9.0 mm
Czech Republic, Southern Bohemia, Slavče
Unequally trihedral, with grooving on some planes, olive green, with some attributes of Muong Nong tektites, 18.5 × 15.0 × 12.0 mm
Moldavites were first cleaned in distilled water in an ultrasonic bath, dried, and weighed. Then, they were etched in analytical quality 5% hydrofluoric acid at 20 °C for 4 h. During this process, surface layer of the moldavite glass, which can by hydrated and secondarily contaminated by carbon during weathering, was dissolved prior to sample crushing and homogenization. Samples were then repeatedly washed with distilled water, dried, and weighed again. The weight loss of the samples during the etching was between 0.74 and 1.11 wt%. This removal of surface layer of the studied samples should be sufficient as the generally extremely low water content of moldavites indicates that glass hydration does not penetrate deeper into the glass. From that moment, samples were always stored in purified glass containers, never touched by either bare hand or any plastic or carbon-containing tools, and treated in a dust-free environment.
The samples were then homogenized in a new, purified porcelain mortar to a grain size of <63 μm. During this process, all gases contained in potentially present bubbles escaped, and the only form of carbon present in the sample was carbon contained in the glass or CO2-gas adsorbed on the surface of pulverized sample. The carbon content of the aliquots of homogenized samples was first tested by an ELTRA CS Analyzer (combustion in O2 carrier gas and infrared detection of CO2), but the samples had carbon content below the detection limit of the apparatus (i.e., below 100 ppm).
Samples were then analyzed using an online flash combustion-gas chromatography-isotope ratio mass spectrometry method (FC-GC-IRMS; Fisons 1108 Element Analyzer—ConFlo interface—Finnigan MAT 251; Laboratories of Czech Geological Survey, Prague). Overall analytical uncertainty of the δ13C value of this method is ± 0.2 ‰, if the sample size is sufficient. Disadvantage of this method is the limited size of a sample, which can be loaded into the analyzer (about 100 mg). The sample is combusted together with a small tin crucible in helium carrier gas with a pulse of oxygen. Oxidation of the tin crucible locally increases the temperature to approximately 1700 °C. After chromatographic separation, CO2 is online introduced into the isotope mass spectrometer. The CO2 yield was low again and indicated a C content in the samples of only approximately 30 ppm. Precise measurement of the carbon isotopic composition of the produced CO2 was impossible. Nevertheless, rough measurement indicated that the δ13C of the produced gas was about −26 ± 2‰ VPDB (VPDB means Vienna-PDB, a common stable isotope standard used for determination of carbon and oxygen isotopic composition, which replaced the exhausted Pee Dee Belemnite standard). Size limits of the tin capsules excluded the measurement of larger sample aliquots.
The only possibility for how to estimate the carbon content and measure the carbon isotopic composition in the moldavite samples more precisely was thus to use an offline combustion/oxidation laboratory line, which would enable loading a larger sample. In the apparatus used (Laboratories of Czech Geological Survey, Prague; constructed on similar principles as the laboratory line described by Dumke et al. 1989), the sample is loaded on a quartz-glass tray and inserted into a quartz reactor with flow of high-purity oxygen carrier gas at a pressure of 1 atm. After sample oxidation at 1000 °C, the gases continue through a furnace with CuO, where potentially present CO is oxidized to CO2. After a capillary reduction of the pressure of carrier gas to approximately 100 Pa, CO2 is cryogenically separated in two subsequent freezing traps, and residual carrier gas is pumped away. The CO2 and H2O are then cryogenically separated from each other, and CO2 transferred into a transportation ampoule. The carbon isotopic measurement is performed using a Finnigan MAT 251 mass spectrometer with a precision of the δ13C value better than ±0.1 ‰.
The reaction yield can be monitored either using a Pirani vacuum gauge on the preparation line, or by peak height at the mass spectrometer, both calibrated by known quantities of carbon. The calibration curve was determined using precisely weighed (±10 μg) small quantities of pure graphite processed in an identical way as the samples, including graphite loads producing signals close to those of samples. The blank of the laboratory preparation line (standard procedure with no sample loaded) is about 7 μg of C. It was determined only by reading on the Pirani gauge. With respect to the method of the carbon yield determination, no blank correction was needed. No blank correction was possible for the determination of the isotopic composition as the δ13C value of the (very small) blank yield was not known and could not be determined with respect to the small gas quantity. The studied moldavite glass samples did not melt during the experiment.
To characterize the studied moldavites chemically, the samples were also analyzed for the contents of 35 elements using instrumental neutron activation analysis (INAA; Table 2). The analytical procedures were similar to those described in Řanda et al. (2008).
To characterize the surface sediments of the target area, 23 samples of Upper Marine Molasse and Upper Freshwater Molasse were collected around the Ries crater. They were analyzed for the contents of carbonate and organic carbon using common commercial methods (Coulometric Titration and Element Analyzer).
Chemical Composition of Glass of the Studied Moldavites
Chemical composition of the studied moldavite samples (see Table 2) is close to an average moldavite composition, which can be best represented by the composition of the bulk of South Bohemian and/or Moravian moldavites (cf. Řanda et al. 2008). Figure 4 shows the contents of the main components of the studied moldavites in comparison with chemical composition of moldavites from three major substrewn fields obtained from the same laboratory using identical procedures. Compositions of samples No. 151 (Moravian) and No. 153 (South Bohemian) are similar to each other, and with somewhat lower Ca and Mg contents and slightly higher Al and Na contents may be closer to the average composition of Moravian rather than South Bohemian moldavites. On the other hand, the contents of other elements, such as Fe and REE, are closer to South Bohemian than Moravian moldavites. Sample No. 152 (South Bohemian) differs from the other two samples and is characterized by markedly higher Ca and Mg and lower Na contents. This partly resembles the composition of some moldavites from the Cheb Basin or the so-called “poisonous green” moldavites—a subgroup of South Bohemian moldavites (see Řanda et al. 2008; Skála et al. 2009).
Carbon Content in Moldavite Glass
Carbon content in moldavite glass is very low, below the detection limit of conventional C/S element analyzer (<100 ppm). Similarly, the online flash combustion-GC-IRMS method indicated a very low carbon content in the three studied samples (approximately 30 ppm or below). The offline oxidation of a large sample quantity in the preparation line yielded CO2 in quantities exceeding the blank of the line by the factor of >10. Carbon content in the studied samples determined using this approach ranges from 35 to 41 ppm C (see Table 3). The sample from the Moravian substrewn field showed a slightly higher carbon content than the two samples from the South Bohemian substrewn field.
Carbon Isotopic Composition of Moldavite Glass
The quantity of CO2 gas produced during the oxidation of moldavite glass in the offline laboratory preparation line was sufficient for a precise measurement of the δ13C value. The δ13C values of the prepared CO2 gas ranged from −28.5 to −29.9‰ VPDB. Data for individual samples are contained in Table 3. These data are, as far as we know, the first published carbon isotope data on moldavites.
Carbon Content of Neogene Sediments of the Target Area
The unconsolidated sediments of the Upper Marine Molasse and the Upper Freshwater Molasse believed to cover the surface of the target area at the time of impact are generally poor in organic matter, but locally rich in clastic carbonate grains. The content of organic carbon in these sands and clays ranges between <0.01 and 0.21 wt% (average 0.09 wt%; 23 samples), while the content of carbonate carbon is in the range from <0.01 to 7.67 wt% (average 1.81 wt%; 23 samples).
The validity of the discussion is limited by the fact that the available three analyses represent only a pilot study. Analysis of a larger moldavite sample set covering also other substrewn fields is needed, including carbon isotope analyses of the gaseous content of bubbles.
There is a general consensus that moldavites represent reworked terrestrial materials from the impact site in the Ries area, with dominance of materials derived from very shallow, near-surface sediments and soils (Engelhardt et al. 1987, 2005; Řanda et al. 2008; this concept of tektite origin was generally first published probably by Schwarcz 1962). Local unconsolidated surface sediments, mostly the Miocene Upper Freshwater Molasse, are quartz-dominated with a variable proportion of carbonate grains. These high-porosity sediments were water-saturated and probably covered by a mostly carbonate-free soil layer and abundant vegetation.
It can be expected that during the impact process water from the near-surface Earth’s environments was dissociated to hydrogen and oxygen, and that the oxidation of abundant organic compounds consumed all free oxygen. Enormous quantities of volatiles evolved from dissociation of water (cf. Kieffer and Simonds 1980; Howard 2011), and the combustion/decomposition of the organic matter could enhance the high-velocity melt ejection. As indicated by the usual absence of nitrogen in the bubbles of tektites, and as it is generally accepted, the original terrestrial atmosphere was displaced by the extreme pressure outside the area where moldavites were formed. If this hypothetical mechanism was active, then the carbon isotopic composition of the CO2 contained in bubbles (and in the glass) of moldavites should be generally between that of sedimentary carbonate (δ13C ∼ 0 ‰ VPDB) and that of terrestrial vegetation and soil, usually dominated by C3 plants (δ13C typically from −33 to −24‰ VPDB), with a smaller proportion of C4 plants (δ13C typically from −16 to −10‰ VPDB; cf. O’Leary 1988).
As it already has been shown based on chemical composition of moldavites (see references above in the General Characteristics of Moldavites and Their Strewn Field section), deeper-seated rock units (Mesozoic and Permian sediments and the underlying Hercynian crystalline basement of the Ries) did not participate in the formation of moldavites. These were derived from the uppermost layer of unconsolidated sediments and soils, including abundant vegetation. Carbon isotope data obtained by Abbott et al. (1996, 1998) and Abbott (2000) for suevites are therefore of little relevance for the discussion on moldavites.
Molassic sediments of the Ries area contain 0.09 wt% organic carbon and 1.81 wt% carbonate carbon on average. In contrast, living vegetation contains about 50% carbon (in dry mass). Simple mixing calculation shows that organic carbon dominates already a mixture consisting of 90% of molassic sediments and 10% of organic matter derived from vegetation, and, of course, in any other mixture containing more than 10% organic matter.
Carbon content determined in our three moldavite samples is significantly lower than that of the Upper Marine and Freshwater Molasses of the target area. This clearly shows that most of the carbon released during moldavite formation must have stayed in gaseous form and was not incorporated into the moldavite. Moldavite δ13C values indicate a dominantly organic source of carbon. There is no indication of incorporation of heavy carbon from limestone to moldavites. Deeper rock sources from the Ries are therefore interpreted as not being incorporated into the moldavite melt. If so, the isotopic fingerprinting from carbonates should be more distinct; this is not the case. These data thus indicate that the dominant source of moldavite melt was the uppermost layer of soils and unconsolidated sediments covered by abundant vegetation, which contained only a limited amount of carbonate.
The low content of volatile compounds in tektites has already been noticed by other authors. Based on the previously published data, O’Keefe (1976) stated that tektites contain usually 50–150 ppm carbon. Petersile et al. (1967) extracted 72 ppm of carbonaceous compounds from crushed moldavite glass. Our data obtained on carefully cleaned moldavite samples indicate somewhat lower carbon contents in moldavite glass, 35–41 ppm. These low contents of volatiles are probably related both to high temperature of moldavite formation and to low external pressure during the glass solidification, where most volatiles stayed in the gas phase outside the moldavite glass. Effective separation of volatiles from moldavite glass is, besides the low carbon content, confirmed also by its low water content. Water content in moldavites is the lowest of all tektites, only 60–100 ppm (Beran and Koeberl 1997).
The carbon isotopic composition of the CO2 produced by high-temperature oxidation of the moldavite glass (−28.5 to −29.9‰ VPDB) falls within the general range of Neogene terrestrial organic matter, dominated by C3 plants. As a geographically and temporarily relatively close equivalent, the carbon isotope data of Miocene coals formed in adjacent parts of Europe can be used. Lücke et al. (1999) found the average δ13C values for Miocene coal components in German Lower Rhine Embayment between −23.3 and −26.0‰ VPDB, Hámor-Vidó and Hámor (2007) measured values from −23.92 to −28.54‰ VPDB in the Panonian Basin Mesozoic to Tertiary coals. Miocene coals of the Most Basin, Czech Republic, average at a δ13C value of −26.0‰ VPDB (average of 117 samples, Mach, personal communication, Bílina Mines). Bechtel et al. (2008) reported δ13C values from −24.6 to −26.7‰ VPDB for Early Eocene to Pliocene coals from the Alpine Realm and Middle German Lignite District. Data obtained on moldavite glass therefore indicate that the main source of carbon during moldavite formation was terrestrial organic matter, and that the carbon derived from decomposition of carbonate rocks was not an important carbon source. This conclusion is in agreement with the interpretation of Řanda et al. (2008) who considered terrestrial organic matter one of the important sources in moldavite formation.
As discussed in the above article, the potential biogenic component of the moldavite source material is possibly indicated by the enhanced K/Na (and K/Rb) ratio. This would be a result of enrichment in elements essential for plants and depletion in nonessential elements, which is even more pronounced for Ca/Sr and Ca/Ba ratios (Burton et al. 1999). The suggestion for organic input is strongly supported in moldavites from the Cheb Basin, which shows extremely elevated K/Na ratios (see Řanda et al.  for the detailed discussion). These ratios are significantly higher in sample No. 152 than in samples No. 151 and 153, although they are considerably lower than the extremely high values observed in some Cheb Basin moldavites and several South Bohemian moldavites (Řanda et al. 2008; Skála et al. 2009). However, these possible indicators of a higher input of organic matter in sample No. 152 do not seem to be reflected in either higher carbon content or different carbon isotopic composition measured in this sample.
In the future, δ13C and δ18O of CO2 contained in bubbles in the moldavites should be determined, using a low-temperature gas extraction by sample crushing to eliminate the influence of C contained in the glass. At the temperatures of moldavite formation above 1000 °C, carbon isotope fractionation between reduced carbon and CO2 is of only a few ‰ (in the isotope equilibrium; cf. Bottinga 1969). Therefore, the isotopic composition of carbon in moldavite glass and that of CO2 in the bubbles should not differ dramatically.
Determination of the δ18O value of CO2 contained in moldavite bubbles can be of a very high importance given the fact that oxygen isotopic compositions in the atmospheric oxygen and in water are largely different. Such data should therefore allow us to discriminate whether the oxygen consumed during the organic matter oxidation was atmospheric oxygen or oxygen from meteoric water dissociated to hydrogen and oxygen. With respect to the usual absence of nitrogen in bubbles of tektites reported in the literature, the latter scenario seems to be more probable.
Bubbles in tektites can be formed by several different mechanisms, including an entrapment of very diluted external atmosphere during tektite glass solidification, a release of gases dissolved in the glass during glass cooling above the temperature of glass transformation, and thermal volumetric contraction inside the still plastic glass, when the surface of the tektite has already solidified.
A review of previously published data has shown that the gas pressure in the bubbles is generally low, and some bubbles contain only nondetectable quantities of gas. Gas pressure in bubbles in the layered tektites was reported to be higher, reaching up to one-third of the atmospheric pressure. Composition of the contained gases published in most articles sharply differs from the composition of the terrestrial atmosphere, with oxygen and nitrogen either absent or present in only minute amounts. Gas in the bubbles is usually dominated by either CO or CO2, and H2. The CO/CO2 ratio seems to depend on the extraction procedure used. When thermal extraction is used (tektite melting under vacuum), the gas yield and CO/CO2 ratio are higher. This is probably related to the mobilization of carbon contained directly in the glass during the thermal extraction. Analyses using low-temperature tektite crushing usually contain only a small CO amount, or none.
Three moldavite samples analyzed within the present study contained only 35–41 ppm of carbon in the glass. Their δ13C values ranged from −28.5 to −29.9‰ VPDB. This indicates that the terrestrial organic matter must have been a dominant carbon source during the formation of moldavites, while the role of carbonates was a subordinate one.
Acknowledgments— This study was financed by the Czech Science Foundation project GA 205/09/0991 and performed within the research program No. AV0Z30130516 of the Institute of Geology AS CR. We wish to thank Prof. Dr. Kurt Heissig for the assistance during sampling of sediments in the Ries area. Comments of the reviewers Dr. Iain Gilmour and Dr. Kieren Torres Howard, and of the associate editor Dr. Natalia Artemieva, significantly improved the manuscript and are highly appreciated.