Proceedings of the XIth International Congress and Exposition June 2-5, 2008 Orlando, Florida USA ©2008 Society for Experimental Mechanics Inc. Obsidian - Natural Nanostructured Glass: Preliminary Research Results B.F. Dorfman, Dr of Sc, Clarkson University, 8 Clarkson Avenue, Potsdam, NY 13699, [email protected]; 229, 19th Ave., #9, San Francisco, CA, 94121; L.C. Zhang, Ph.D., Clarkson University; C.E. Skinner, Ph.D., Northwest Research Obsidian Studies Laboratory; S.C. Zunjarrao, R.P. Singh, Ph.D., Oklahoma State University ABSTRACT Correlated Mechanical-Structural-Chemical Analysis of obsidians (from 15 American localities) by micro- and nano-indentation, XRD, SEM, TEM (2-5 nm resolution), SEM-EDS, XRF revealed distinguished microfracture behavior of volcanic glass vs. artificial glasses, fused quartz, and nano-crystalline silica, as well as fused homogenized obsidians of the same origin, while the crack propagation threshold in obsidian is superior to these reference materials in order of magnitude, and the fracture pattern in obsidian is predominantly two-dimensional (2D) superficial vs. quasi one-dimensional (1D), deep radial cracks dominating in artificial glass. Compositional profiling of obsidian detects sharp change in relative content of oxygen across inter-domain borders and across deflected micro-cracks thus revealing correlation ‘oxygen distribution - domain structure - fractural mechanics’. It suggests that relatively slight structural-chemical variation resulting with hierarchical network of inter-domain borders may produce essential mechanical reinforcement of glass. If proofed, the last conclusion represents particular interest for glass industry. Beyond of that, the obsidian formation covers a gap between geological and technological timescales; its study may lead to new findings useful for various fields of materials development. However, due to extreme complexity of obsidian structure and mechanisms of formation, the further research needs cross-disciplinary efforts. 1. Introduction Obsidian as natural volcanic glass is commonly assumed possessing a fracture behavior identical to conventional artificial glass albeit it employed from the pre-historic tool and arm and up to contemporary surgical tool in the manner, which would be unfeasible for various artificial glasses. In spite of some early remarkable researches [1-4], the prevalent present definition of obsidian is not strongly differentiated from a definition given over decades ago and structurally equalizing volcanic glass with common glass. Of about 3200 publications on obsidian referred in the literature up to 1998 [5], only 8 papers [1-4, 6-9] concern with mechanical properties of natural glasses. Although archeologists and anthropologists were aware about the clearly distinguishing mechanics of obsidian long ago [2], correlation between such a unique mechanics and specific structural or compositional features of volcanic glasses remains unknown. Recent researches in the obsidian’s structure and composition are extensive (see some examples: [10-14]), but predominantly associated with geological knowledge of its formation, devitrification and age, as well as archeological tracing with rather “axiomatic” acceptance of its unique mechanical properties underlying its ancient use and archeological significance. This article is narrowly focused in search for structural/chemical features which may be responsible for mechanical properties generally distinguishing volcanic glasses from artificial glasses. Any analysis of obsidian may encounter the problem of extreme complexity: variation of basic chemical composition of glass matrix combined with diversity of crystalline inclusions depend in turn on geophysical and thermochemical conditions of formation and cooling of volcanic mass. To make the task at least in principle solvable, we selected certain set of samples assuming that they represent sufficiently broad diversity of these volcanic materials but avoiding any analysis of actual mechanisms of those rocks formation. Certainly, as soon as the responsible features will be defined, the respective analysis of their genesis will be absolutely necessary. 2. Experimental Brief description of 15 examined samples is shown in Table 1, including the samples’ origin, geological age and content of bulk elements. Since the trace elements are generally used for identification of the obsidian’s sources [15], concentration of Fe, Ti, Mn, Zn, Ga, Rb, Sr, Y, Zr, Nb, Ba, Pb, Th was analyzed by the XRF and exhibited in Table 2 for all the examined samples from various localities, as shown in Table 1. Numbers of standard deviations are also indicated. The tables mark all the samples with numbers, which used in the following description of experimental results and discussion. Some of samples like #6, #12 and #13 display Table 1. List of examined samples and Overall Chemical contents (in at. %) of bulk elements # Origin Age O Si Al Na K 1 Beatys Butte: OR 10.36 ±0.53 my 47.5 ±5.3 40.5 ±4.5 6.1 ±1.3 3.1 ±1.0 2.8 ±0.7 (K-Ar) 2 Big Obsidian Flow: OR 1,300 50.5 ±5.0 35.3 ±3.5 7.2 ±1.4 4.1 ±1.3 2.9 ±0.8 yrs BP 3 Burns Butte:OR 7.54my (K-Ar) 56.0 ±6.4 29.8 ±3.3 7.3 ±1.5 4.1 ±0.9 2.8 ±0.6 4 Cougar Mountain: OR 4.31±0.3my 51.4 ±5.2 35.0 ±3.2 7.4 ±1.0 3.5 ±0.8 2.7 ±0.3 5 Glass Buttes V.1: OR 4.91±0.7my 54.5 ±4.3 30.5 ±2.4 7.9 ±1.3 3.7 ±0.7 3.4 ±0.5 6 Glass Buttes V. 3:OR 46.0 ±3.5 40.0 ±3.3 6.7 ±0.8 3.6 ±0.6 3.7 ±0.3 7 Glass Mountain: CA 900 yrs BP 55.0 ±4.6 30.2 ±3.5 6.9 ±1.0 4.4 ±0.8 3.5 ±0.5 8 Horse Mountain: OR 6.9±0.14 my 57.6 ±5.3 28.5 ±3.4 6.9 ±1.3 4.2 ±0.5 2.8 ±0.3 9 Mckay Butte: OR. 0.58±0.10 my 44.3 ±3.3 40.7 ±3.1 7.4 ±0.9 3.4 ±0.4 4.2 ±0.3 10 Interlake Obs. Flow:OR 7,300 yrs. BP 43.6 ±4.5 40.8 ±4.3 7.8 ±1.1 4.3 ±0.7 3.5 ±0.5 11 Glass Buttes 47.4 ±4.7 38.5 ±3.3 7.5 ±0.8 3.5 ±0.3 3.1 ±0.3 12 Glass Buttes 6 46.5 ±4.9 39.3 ±4.4 7.6 ±0.7 3.4 ±0.3 3.2 ±0.3 13 Buck Mountain, N. CA 46.1 ±5.0 38.8 ±3.3 7.7 ±1.0 3.9 ±0.6 3.5 ±0.5 14 Brazil. Details unknown 37.5 ±4.3 48.8 ±5.0 6.5 ±0.7 3.4 ±0.4 3.8 ±0.5 15 Mexico. (“Rainbow”) 42.4 ±4.1 43.7 ±4.2 7.6 ±0.8 3.7 ±0.3 2.6 ±0.2 Table 2 Overall Chemical compositions. Trace elements (min/max, unit: ppm except Fe in wt.%) ## Ti Mn Fe Zn Ga Rb Sr Y Zr Nb Ba Pb Th 1 728/ 349/ 0.97/ 24/ 13/ 123/ 165/ 13/ 159/ 10/ 863/ 16/ 4/ 993 442 1.24 41 18 132 176 19 164 13 1021 23 10 2 1149/ 425/ 2.03/ 57/ 16/ 118/ 53/ 44/ 353/ 20/ 776/ 13/ 9/ 1389 514 2.40 70 21 127 60 48 367 24 882 21 14 3 908/ 234/ 1.20/ 43/ 16/ 120/ 26/ 39/ 244/ 28/ 475/ 17/ 8/ 1253 349 1.69 60 21 131 30 44 274 31 605 22 15 4 250/ 226/ 0.76/ 68/ 14/ 92/ 36/ 52/ 127/ 10/ 1123/ 15/ 4/ 390 346 1.20 85 22 102 41 57 140 14 1377 19 13 5 301/ 251/ 0.67/ 19/ 10/ 65/ 22/ 42/ 90/ 9/ 1106/ 8/ 1/ 621 414 0.84 44 25 82 49 50 117 15 1225 19 13 6 461/ 237/ 0.64/ 30/ 11/ 99/ 66/ 25/ 109/ 6/ 1108/ 15/ 9/ 628 328 0.97 39 25 108 72 29 115 11 1273 20 12 7 1327/ 209/ 1.47/ 37/ 15/ 146/ 107/ 25/ 219/ 7/ 719/ 19/ 12/ 1844 280 1.90 45 21 163 126 28 234 14 773 25 22 8 801/ 437/ 2.81/ 177/ 13/ 128/ ND/ 100/ 618/ 34/ 42/ 33/ 7/ 1073 593 3.23 205 25 143 5 106 640 41 62 40 16 9 833/ 258/ 1.41/ 54/ 17/ 127/ 55/ 37/ 202/ 7/ 968/ 18/ 9/ 1077 436 1.76 117 27 139 62 42 214 15 1072 24 20 10 1150/ 323/ 1.76/ 43/ 43/ 133/ 60/ 40/ 282/ 16/ 766/ 17/ 8/ 1350 372 2.03 60 60 141 66 45 290 21 843 21 17 11 Assumed same as #5 12 501 608 0.71 23 21 108 69 18 99 9 644 21 5 13 556 327 0.74 36 15 111 62 24 98 8 1004 20 10 14 596 770 0.77 52 20 184 24 26 131 37 51 30 16 15 482 252 1.42 114 25 159 10 61 510 59 27 16 9 ER ~90 ~35 0.11 6-7 2-3 2-3 6-7 ~2 6-7 1-2 ~20 ~3 ~3 the bi-color or multi-color features. For purpose of the sample identification, analysis of trace elements (Fe, Ti, Mn, Zn, Ga, Rb, Sr, Y, Zr, Nb, Ba, Pb, Th) in obsidian was conducted using a Spectrace 5000 energy dispersive X-ray fluorescence (XRF) spectrometer. The system is equipped with a Si(Li) detector with a resolution of 155 eV 2 FHWM for 5.9 keV X-rays (at 1000 counts per second) in an area 30 mm . Signals from the spectrometer are amplified and filtered by a time variant pulse processor and sent to a 100 MHZ Wilkinson type analog-to-digital converter. The X-ray tube employed is a Bremsstrahlung type, with a rhodium target and 5 mil Be window. The tube is driven by a 50 kV 1 mA high voltage power supply, providing a voltage range of 4 to 50 kV. The overall structures of the samples were analyzed by obtaining X-ray diffraction (XRD) patterns from Bruker D8-02 X-ray diffractometer. Morphologies of obsidians were characterized by JEOL JSM-6300 scanning electron microscope (SEM) and JEOL JEM-2010 Scanning transmission electron microscope (TEM). The SEM samples were firstly grinded and polished. In order to reveal the real morphology of the inclusions hidden within the obsidians, some areas of SEM samples were etched by 10vol.% HF solution. To avoid the electrical charge effects, the thin coating of vaporized Au had been applied. For TEM sample preparation, the broken pieces of samples were grinded into fine powders. A suspension containing these fine powders was made in an ethanol medium. A drop of the suspension was deposited on to carbon coated TEM grid and dried under vacuum. The analyses of the bulk elements Si, O, Al, Na and K in obsidian were carried out by means of the same SEM served with energy dispersive spectroscopy (EDS) systems capable of detecting light elements down to boron. The analyses in this research include: overall chemical analysis; local micro-chemical analysis for identifying various inclusions; a linear EDS scanning, in particularly cross the internal borders of obsidian structures. The SEM-EDS microanalyses can give rather comparable data. The errors due to different surface geometry can be negligible because of very small and practically flat surface of the analyzed samples. Worth to note, determination of the oxygen concentration is still challenging since oxygen generates very soft X-ray subject to strong absorption by either the surface or the detector window. To assure convincing results, the SEM-EDS microanalysis on the uncoated (less-conductive) samples was also performed with comparisons between the corresponding EDS spectra. For standard quantification, the fused quartz, i.e. stoichiometric SiO2 (of semiconductor purity) was used as the standard reference sample. The comparative EDS spectra were collected from the obsidian samples and standard fused quartz, with all of them simultaneously mounted on the SEM sample stage in order to maintain the similar beam and surface conditions for a series of spectra collections. Combined nano- & micro- indentation covering the load range from a few mN to 24 N (except the gap 100<L<150mN) was found as the most effective for obsidian study. Nanoindentation had been conducted with standard nanoindenter in the following conditions: Load controlled experiments; Indentation using a diamond Berkovich tip; Peak load of 100 mN; Poisson’s ratio assumed to be 0.3. Samples were cut by high speed tile saw, mounted in epoxy and polished to obtain a mirror finish before indentation. Table 3 Nano-indentation on Obsidian (GPa): Sample E H ∆E ∆H Fig. 1 Typical Load-Displacement plots for nanoindentation of obsidians 1, 12, 13 1 67.307 0.217 7.849 0.138 12/Red 12/black 13/red 69.123 69.795 67.423 0.688 0.880 0.696 8.662 8.8 9.003 0.049 0.172 0.356 13/pink 65.832 0.723 8.576 0.034 A standard manually operated Vickers microindenter was employed in manual mode allowing investigation of fracture behavior in exceptionally wide range of available loads and convenient direct observation of post-load fracture phenomena. Maximum loads up to 24.0 N were employed in this research. Because such extraordinary loads could possibly damage the diamond pyramid, the indenter was periodically examined and calibrated by the hardness measurements of reference samples (aluminum, steel, glass and single crystal silicon). For a comparison, conventional lime glass (“window glass”), Hv= ~5.5 GPa; optical polished glass, Hv=~6.0 GPa, and fused quartz (synthetic polished substrate of semiconductor industry quality), Hv=~9.0 GPa had been examined with the same technique. 2b 2a 2b 3b 3a 4a 5a 4b 5b Fig.2 Typical fracture pattern in Obsidian (a) at L=13N and in Lime glass at 12N (b) Fig.3 a,b. Typical fracture pattern in Lime glass at Loads 16N (a) and 24N (b) Fig.4 Obsidian. L=16N. 4a Typical fracture pattern. 4b EDS profile across the deflected crack Fig.5. Obsidian. L=24N 4a Typical fracture pattern 5b EDS Compositional profile across the deflected crack In addition, three kinds of obsidian were fused using CARLISLE bench burner up to clear, homogenized, virtually stress-free state. Two fusing methods tested: Low temperature fusing with propane/oxygen 0 flame at ~1000 C and high temperature fusing with hydrogen/oxygen 0 flame at ~1500 C. At 0 ~1000 C, a complete homogenization reached an essentially longer fusing time (up to about 1 hour total time of multistep fusing and annealing), but the final material is virtually free of gas bubbles. In the higher temperature range, the optically homogeneous state may be produced during a few minutes, but an extensive gas emission results with microbubbles formation, and even more essential time is required to produce uniform, clear, stressfree glass. A simple polaroscope was used to check the residual thermally generated stress. The mechanical behavior of homogenized fused obsidians was evaluated qualitatively. 3b 3. Experimental results Typical results of nano- and micro-indentation shown on Fig. 1-5 and Table 3. It was found that the fracture of obsidians (Figures 2a, 4, 5) is strongly differentiated from both common glasses and fused quartz. The differences are reflected in four aspects as follows: 1) Fracture pattern: Up to extremely high loads of 13-15 N, the fracture in obsidian is predominantly two-dimensional (2D) superficial vs. quasi onedimensional (1D), deep radial cracks dominating in artificial glass (Fig. 2b, 3 a, b). In detail, 2D cracks in obsidian usually propagate on ≤ 100µm level only and then arrested by certain micro-irregularities. Alternatively in glass, the 1D cracks nucleate at loads <<1.0 N and propagate in radial directions from indent, evidently becoming deeper with the load increase. 2) Fracture threshold: Two subsequent thresholds were found in obsidians: a) ~4.2N (mean value) for occurrence of superficial micro-chipping. No cracks propagation beyond of crackarresting stressed area found below or beyond of this threshold and up to the second threshold. b) ~14N for radial crack propagation – over an order of magnitude exceeding that of artificial glass (~0.6N). 3) Microhardness: A complex micro-hardness behavior of obsidian revealed over the entire range of loads: at loads < ~1N, the Vickers micro-hardness approximately corresponds to mineralogical data 5.0– 5.5 GPa; at the higher loads ≥ ~2N, the Vickers technique shows two differentiated features: superficial imprint corresponding to Hv=5-6 GPa, and deep imprint leading to hardness of Hv=6.5-9.5 GPa, thus approaching the quartz values. Accordingly, the measured Hv values increase with the load increase, contrary to the normal behavior of glass. 4) Fracture toughness: Based on the crack indentation pattern by the Lawn or Pharr methods [16, 17], the fracture toughness of obsidian may be evaluated as Kk > 1.6 - 3.9 MPa√m. For a comparison, the fracture toughness of artificial glass and fused quartz with the same indenter was also evaluated: a) Polished optical glass: Kk = 0.84 MPa√m, which reasonably corresponds to referred value Kk = 0.70 MPa√m [18]. b) Fused quartz: Kk > 0.64 MPa√m which corresponds to Kk = 0.58 MPa√m [18]. Fracture toughness of obsidian essentially exceeds the values for polished optical glass and fused quartz. Distinctive mechanics of obsidian implies existence of some topological hierarchy in its structure. In this research, the structures, intrinsic morphologies and compositional distributions in various obsidians had been examined from over 10 certified locations in the USA, as well as two random samples from Mexico and Brasilia. Certain intrinsic particularities of obsidian, which may be responsible for its unique mechanics, are considered below. Typically, obsidians are rich with inclusions, and their possible contributions in obsidian mechanics must be clarified first. Figures 6-9 show represent a brief summary of examined varieties. Most inclusions (Fig. 6,7) identified as the silica phases (such as quartz, cristobalite and moganite) and the silicate phases (like sodium silicate), although some minor inclusions of complex silicate were not determined yet. The degree of crystallinity in the a-matrix can be roughly estimated in term of the respective heights of XRD peaks from the a-matrix and crystalline inclusions. Accordingly, the obsidians are arbitrarily separated into three categories: (1) Fig. 6(a): six with low population of inclusions; (2) Fig. 6(b): six with medium population of inclusions; (3) Fig. 6(c): three with high population of inclusions (note: the only purpose of this classification is analysis of possible correlation ‘structure vs. mechanics’; general classification of structure and inclusions of obsidian is out of scope of this research). A series of SEM images in Fig. 6 which were obtained from etched obsidians, reveal the abundant features regarding both a-matrix and inclusions, including frequent morphological evidence of self-organizing phenomena. Figs. 7 (a)-(c) exhibit common features of the nearly semi-transparent obsidians (like the smoky #14, the jet-black #11 and the smoky area of the red/black #12). Fig. 7(a) displays small inclusions with typical size of <0.2µm. The small inclusions are randomly distributed. Besides, some of inclusions display semi-clustering patterns (see Fig. 7(b)). Some snowflake inclusions of alkali chloride crystals were also observed, as displayed in Fig. 7(c). Figs. 7(d)-(f) are typical images from the black and dark-gray obsidians. Fig. 7(d) displays a high-density of large inclusion clusters in the Rainbow #15. These large clusters with random distribution are virtually selforganized structure in the a-matrix. In other black obsidians, the similar clusters but in lower populations were observed, as illustrated in Fig. 7(e). In addition, the rod-like inclusions, usually co-oriented in one direction often found in the obsidian #7 - Fig. 7(f) give an example (chemical composition - Fe, Mg and Ca-rich – shown below on Fig 9(d)). Figs. 7(g)-(l) are some SEM images from the red area of the black/red #12, the brown #6 or the pink #13. Figs. 7(g), (h) and (j) reveal typical lined-up feature of large inclusion clusters, and Fig. 7(i) displays some of clusters adopting the snowflake pattern. Fig. 7(k) shows the hierarchical structure of one large cluster and different scales in its hierarchical arrangement. Fig. 7(l) shows groups of clusters, each consisting of even smaller inclusions in turn. TEM observations further confirmed the hierarchical structure of inclusions down to several nanometers in size. Fig. 8 shows an example: each inclusion embodied in one of numerous clusters detectible by SEM is explored by TEM as actually a hierarchical cluster as well, embracing several of even smaller particles. Chemistry analysis of obsidians was carried out through the SEM-EDS method. The standard fused quartz was simultaneously inserted with the studied obsidians. The overall EDS spectra were collected over very large area of samples, so reliable mean chemical compositions could be obtained from the analyzed samples with inhomogeneous features. Fig. 9(a) shows an overall EDS spectrum of the rainbow #15 in comparison with that of fused quartz standard. It is clearly seen that the O:Si ratio in peak heights as well as peak integral intensities of the rainbow #15 is nearly half of that of fused quartz, implying the severe oxygen-deficit of silica in the rainbow #15. A series of similar results of comparative overall EDS spectra were also obtained from other obsidian samples. Table 1 summarizes the average concentrations of bulk elements Si, O, Al, Na and K respectively according to standard quantification of more than 5 overall EDS spectra from each individual kind of obsidians. All of the EDS results in Table 1 display that almost all of the 15 obsidians (except of #3, #8) are in different degrees of oxygen- deficit with the lower O:Si ratios than that of fused quartz, whereas the other bulk elements - Al, Na, K Fig. 6 Series of XRD patterns in the range of 19°-39° for all the 15 obsidians. In the XRD patterns, the broadening peaks between 19° and 29° originate from the amorphous matrix (a-matrix), and the individual sharp peaks from crystalline inclusions in a-matrix #1-15 classified in three groups: (a) low population of inclusions; (b) medium population of inclusions; (c) high population of inclusions. Notes: 1) label q - quartz, c - the cristobalite; sthe sodium silicate phases. Note: this classification is only for illustration of varieties of specimen in this work. Fig. 7 SEM Images from obsidians after etching: (a), (b), (c) semi-transparent obsidians: small inclusions; snowflakes; (d), (e), (f) -black or dark-grey color obsidians: large inclusion clusters;(g), (h), (i) – brown, red or pink obsidians: lined-up of large inclusion clusters; snowflake clusters; (j), (k), (l) - brown, red, pink obsidians: semiordered inclusions and hierarchical structure on a large scale. Note: “Snowflakes” on Fig. 7(c) is not silica devitrification but chloride crystals, apparently, formed during lava solidification 100nm Fig 8 TEM. Obsidian #12. - in different obsidians are at the similar concentration levels. Worth to note, the absolute oxygen concentration by SEM-EDS may need further verification by more precise analytical methods [19], but data for relative variation, especially in the O:Si ratios between obsidian samples and standard fused quartz, may be considered as reliable. The values of O/Si ratio fluctuate even between structural domains of the same obsidians, as demonstrated in SEM-EDS line-scanning analysis of Fig. 9(b). SEM image in the left reveals feather-like domains in dark contrast. Fig. 9 (a) Comparison of the overall EDS spectra between the fused quartz standard and the rainbow #15; (b) The SEM-EDS line-scanning analysis of the pink #13: Left SEM image reveals domains in various contrasts; Right-top - the SEM image with a scanning line across the dark-contrast domain; Right-bottom - profiles of the counts of Si and O during line scanning. (c) The EDS spotanalysis of the black #11: Spot 2 – a-matrix; Spots 1, 3 and 4 – inclusions (with Au coating to avoid the charge effect) (d) The EDS spot-analysis of the obsidian #7: Spot 1 - a-matrix; Spot 2,4 the needle-like inclusions of MgCa-Fe-rich sodium silicate phase. Fig. 9a,b clearly reveals a sharp change of the O:Si ratio in counts across the intrinsic border. This is a common feature for all examined silica matrices where the contrast between certain structural domains is observable. Apparently, this phenomenon is general chemical-structural feature of obsidian. The comparable chemistry of the amatrix and the inclusions in obsidians was also investigated by SEM-EDS spot analysis. Although the inclusions exhibit varieties in different obsidians, they can be basically classified into two types in st term of their chemical analysis: (1) The 1 type of crystalline inclusions (most of the observed inclusions virtually in every obsidian) is the silica or silicate phase in the oxygen-deficit state compared with the a-matrix. Such an example can be seen in Fig. 9(c) obtained from the black #11. It displays that the inclusions in spots 1,3,4 have similar chemistry as the a-matrix in spot 2, but the values of the O:Si ratio for inclusions are dramatically reduced nd to about half of that for the a-matrix. (2) The 2 type of crystalline inclusions (some inclusions adopted, but varied in different obsidians) is so-called ‘foreign inclusion’ containing high concentrations of the foreign elements. The “foreign inclusion’ could be the Fe, Ca, Mg-rich inclusions as shown in Fig. 7(f) or alkali chloride as shown in Fig. 9(c). Such an example of chemistry analysis can be seen in Fig.9 (d) obtained from the obsidian #7. It displays that the needle or rod-shaped inclusions in spots 2, 4 are rich in the foreign elements like Fe, Mg and Ca in comparison with the a-matrix in spot 1. 4. Discussion It is important to point out prior to this discussion: all the presented results concerning the structure, morphology and chemical composition of examined specimens of volcanic glasses are considered strictly at the aspect of possible influence of the respective features on glass’ mechanics. Correspondingly, it is not assumed that any revealed feature has independent significance until its mechanical influence is proved, and on the contrary: features that are proved not to be significant for glass mechanics removed from the further consideration. Initially, we considered two principle classes of particularities of obsidian which may be responsible for mechanic behavior differentiating it from artificial glasses: 1. Innate properties of silica matrix, i.e. chemical variations and/or structural features; 2. Foreign inclusions. Foreign inclusions. The foreign inclusions taking alone cannot be responsible for all revealed mechanical particularities of obsidian, but at least in principle, they can work as crack-arresting features thus resulting with higher threshold values of observable fractural phenomena and their predominantly 2D topology. Therefore, such a possibility should be verified and proved or excluded first. The analyzed 15 kinds of obsidians expressed 2 broadly varied density, structure, chemistry, scale (from ~10nm to ~10 µm grain size) and distribution patterns of foreign inclusions (Fig. 6-9) thus providing sufficiently strong base for a reliable verification. In fact, the characteristic fracture behavior and pattern as described in the introduction were preserved in all of the examined obsidians irrespectively to specific forms of inclusions (Figs. 7 and Fig. 9 show examples), while their density varied from very high to nearly zero, i.e. some of examined obsidians were virtually free from foreign inclusions observable by SEM and TEM techniques (the featureless images are not shown). Furthermore, both nano- and micro- indentation in good consistency show that basic mechanical features of volcanic glasses have no correlation with foreign inclusions (with possible exception for relatively small deviations in E and H values). In the other words, the basic quantitative features of obsidians do not show any dependence on foreign inclusions. A possible role of crystalline silica inclusions cannot be completely excluded; however, the nano-crystalline silica (not shown here) expresses extreme brittleness with very low fracture threshold. Although the possibility of slight dependence of specific values of mechanical properties on micro- and nano-crystals incorporated in silica matrix should not be excluded, it is unlikely that the crystalline inclusions is common responsible for characteristic mechanics of volcanic glass, and the further discussion will be limited with the intrinsic properties of matrix itself. Innate properties of silica matrix. Nanoindentation shows the values of elastic modules of obsidian slightly inferior to glass; hence, the E value cannot be responsible for the unique mechanics of volcanic glass. The higher value of hardness cannot be responsible as well: quartz is harder than obsidian, but demonstrates much lower fracture threshold. Hence, it is unlikely that the mean bulk characteristics are the cause of specific fractural behavior of volcanic glasses. Basic chemical composition and related structural features of silica matrix as a probable cause is discussed below. The content of both trace and bulk elements shown in Tables 2 and 4 basically correspond to well known reference data for volcanic glasses, except that the average O:Si ratios in concentration were found essentially lower than the stoichiometric value of SiO2, and this ratio is essentially and sometimes sharply varied even inside the silica a-matrix of a given sample. In contrast, variation of other bulk elements such as Al, Na and K is slight (except of the foreign inclusions). Basically, the oxygen deficient centers and clusters are very common even in crystalline quartz, including crisotbalite in rhyolite and obsidian [20]. Indeed, formation of absolutely uniform stoichiometric SiO2 during the magma solidification would be thermodynamically improbable. In nonstoichiometric SiOx (x<2), the entire range of O:Si ratio between 0:1 to 2:1 is possible, but most stable forms are apparently grouped in three ranges in vicinities of stoichiometric values: at 0<x<<1; at x1, and at x2. Practically, non-stoichiometric SiOx in proximity of x=1 is commonly used in electronics over half a century, and more recently extensively developed for optimization of Si/SiO2 structures in the course of nano-scaling of semiconductor devices [21-23]. Metallurgically, non-stoichiometric SiOx (x<2), particularly in combination with alumina and other oxides, is able to produce numerous structures differentiating in glass transition temperature, viscosity and diffusion coefficients as well as solid mechanics. In summary, this systematic study revealed two common features for all of the 15 obsidians: (1) Structurally, the obsidians contain many hierarchical levels of structural arrangement: from a few hundreds microns and up to several nm level (See SEM and TEM images in Figs. 7-9); (2) Chemically, all the examined obsidians are characterized with relatively narrow range of aluminum concentrations (about 7.0±0.9 at.%; Table 1) but generally oxygen deficient vs. pure fused quartz, and the O:Si ratio essentially varied cross the observable intrinsic borders (like those separating various domains in Fig. 9(b) or separating inclusion and a-matrix in Fig. 9(c)). Except of one sample shown on Fig. 7d and selected fields of a few more samples where the density of foreign inclusions is too high and they mask the intrinsic obsidian structures, the domain are detectable in nearly all specimens by contrast (as on Fig. 9) or by the closed contours of certain inter-domain borders (Fig. 7 g, j,k,l are the examples). The exact nature of the domains is not finally identified, but it correlation with oxygen content is evident (Fig. 9b gives an example). In most cases, there is a clearly detectable hierarchical arrangement, in some instances fractal-like pattern of the domains; it may be observed by changing the scale of observation (the sequence shown of Figures 7h7l9 gives an example; in fact, many steps of magnification had been used to observe and actually revealed such hierarchical arrangement) or observed even in one image obtained from selectively etched obsidian surface (Fig. 7k). Apparently, this structural hierarchy is due to complex, stepwise post-volcanic processes occurring during the lava’ cooling. As soon as the critical importance of this domain structure for the obsidian mechanics will be proved, the analysis of its correlation with geological circumstances of obsidian formation will be the necessary following step. The fractal geometry exists in numerous natural and synthetic materials [24]. The fractal structures in natural rocks [25] and sandstones [26], manmade glass-ceramics [27] and polymers [28] were also studied by methods of small-angle scattering and SEM, and moreover, fractal materials science has recently became a new promising direction in the field of materials science and engineering. It is likely that such a hierarchical (fractal) structure in obsidians is responsible for their unique fracture pattern since this kind of structure seems to be particularly effective in arresting or deflecting the propagating micro-cracks. So far, some preliminary observations support this assumption; however much more research is still needed to prove it. The nature of this domain structure is not defined yet. The concept of coexisting structural units with fixed Si/Al or even individual Si/Al ratio [29] may suggest a clue, especially combined with the fact that some properties of silica-alumina glasses and melts can show minima or maxima in relationships to Al/(Al+Si) as well as pronounced minimum or maximum values in certain range of the nonbridging oxygen per tetrahedrally coordinated cations ratio (NBO/Si+Al); besides, some narrow range of aluminum content may correspond to the maximum values of specific properties of glass [30, 31]; this may well match our observation that the fused obsidian does not express any mechanical particularities of volcanic glass, even though the only small portion of aluminum lost as the result of fusing. On the other hand, the domain structure is completely dissolving under such high temperature homogenization, and this may be the key reason of drastic mechanical change. The evidence of certain structural domains and correspondingly – inter-domain borders in obsidian is definite, and their crucial influence on mechanical properties of volcanic glass is possible. 5. Conclusion Preliminary examination of micro-mechanical features of obsidian reveals its strongly distinctive behavior vs. various kinds of artificial glass, including common glass, fused quartz, as well as artificial glass produced by fusing the same obsidian. In this research, characterizations of 15 kinds of obsidian by XRD, SEM, TEM, EDS and XRF methods were carried out to define plausible cause of unique mechanics of volcanic glasses. Structures and morphologies from various obsidians are varied, but in many cases the observable intrinsic features are the internal borders dividing areas of certain inborn segregation, at least in some samples clearly hierarchical on 3 to 5 levels, from a few millimeters or larger areas and up to clusters consisting of 3 or 4 nanograins of 2-5 nm size each. While the aluminum concentrations is only narrowly varied from sample to samples and rather uniform in each of them, the O:Si ratio is generally lower than that of stoichiometric SiO2, and it is often varied even in single sample of obsidian. In all samples scanned by SEM EDS technique across the deflected micro-cracks, sharp change of the O:Si ratio was detected. The hierarchy of intrinsically inhomogeneous structure of obsidian, associated with segregation of its major chemical components O/Si at the moderate level of aluminum content, seems to have an explicit correlation with the unique mechanical particularities of obsidian. Being proved by the further research, this would means that a weak domain structure produces strong mechanical reinforcement of volcanic glass - a possible prototype for novel artificial glass. Acknowledgements At the final phase, this research was supported in part with NYSTAR. It is pleasure to express our appreciation to Professor S.V. Babu for his inspiring encouragement. Authors are thankful to Mr. Brian Doran for help in experiments with fusing obsidian. References 1. Sellers G.E. Observations on Stone-Chipping. Annual Report of Smithsonian Institution. 871-891. 1885 2. MacCurdy G.G. The Obsidian Razor of the Aztecs. American Anthropologist (New Series) 2: 417-421. 1900 3. Clark B.R. Stress-controlled orientation of Microlites in obsidian. EOS 51(4): 425-430. 1970 4. Schousboe R. Microscopic edge structures and micro-fractures on obsidian. Lithic Technol. 6(1-2): 14-21 1977 5. Skinner C.E. Obsidian: An Interdisciplinary Bibliography, Northwest Res. Obsidian Studies Laboratory. 1998 6. Abbott P.L., Peterson G.L. Effects of abrasion durability on conglomerate clast populations: examples from Cretaceous and Eocene conglomerates of San Diego area. CA. J. Sedimentary Petrology 48(1):31-42. 1978 7. Boguslavskiy M., Genkin Y. Elastic properties of certain gems. Vyssh. Uchebn. Zaved., Izv., Geol. Razved 8:170-171. 1978 8. Matsushima S. Elastic properties of volcanic glass, glassy rocks and crystalline rocks to 1000°C, 20 Kbar. Int. Geodynamics Conf. W. Pacific and Magma Genesis. Sci. Council of Japan: Tokyo, Japan, pp. 296-301. 1978. 9. Friedman M., Handin J., Higgs N.G., Lantz J.R. Strength and ductility of four dry igneous rocks at low pressures and temperatures to partial melting. Proceedings, Symposium on Rock Mechanics 20:35-50. 1979 10. Okuno M, Iwatsuki H, Matsumoto T. Structural analysis of an obsidian by X-ray diffraction method. Eur J Mineral 8: 1257-1264. 1996 11. Zotov N. Structure of natural volcanic glasses: diffraction versus spectroscopic perspective. Journal of NonCrystalline Solids 323: 1-6. 2003 12. Acquafredda P, Paglionico A. SEM-EDS microanalysis of microphenocrysts of Mediterranean obsidians: a preliminary approach to source discrimination. Eur J Mineral 16: 419-429. 2004 13. Kahle A, Winkler B, Radulescu A, Schreuer J. Small-angle neutron scattering study of volcanic rocks. Eur J Mineral 16: 407–417. 2004 14. Kovacskis V, Dodony I, Labar J R. Amorphous and partly ordered structures in SiO2 rich volcanic glasses, An ED study. Eur J Mineral 18: 745–752. 2006 15 Key C. A. Trace element identification of the source of obsidian in an archaeological site in New Guinea. Nature 219: 360-361. 1968 nd 16. Lawn B. Fracture of Brittle Solids (2 ed.), Cambridge University Press, Cambridge, UK. 1993 17. Pharr G M, Harding D S, Oliver W C. Measurements of fracture toughness in thin films and small volumes using nanoindentation methods. In Mechanical Properties and Deformation Behavior Materials Having Ultra-Fine Microstructures; Nastasi M, Parkin D M and Gleiter H eds.; Kluwer Academic Publishers, Dordrecht, the Netherlands, pp 449-461. 1993 18. Bhushan B. Nanomechanical properties of solid surfaces and thin films. in Handbook of Micro/Nano Tribology, Bharat Bhushan ed., CRC Press, Boca Raton, pp 321-396. 1995 19. Nash W P Analysis of oxygen with the electron Microprobe: Applications to hydrated glass and minerals. American Mineralogist 77: 453-457. 1992 20. Kayama M, Okumura T, Nishido H, Ninagawa K. Cathodoluminescence of natural tridymites and cristobalites. 9th Gen Meeting of the Int mineralogical Association Kobe, Japan. 2006 21. Nicklaw C J, Pagey M P, Pantelides S T, Fleetwood D M, Schrimpf R D, Galloway K F, Wittig J E, Howard B M, Taw E, McNeil W H, Conley J J F. Defects and nanocrystals generated by Si implantation into a-SiO2. IEEE Transactions on Nuclear Science 47: 2269-2275. 2000 22. Zhang R Q, Chu T S, Cheung H F, Wang N, Lee S T. High reactivity of silicon suboxide clusters. Phys Rev B 64: 113304-113308. 2001 23. Tomozeiu N, Faassen E E V, Arnoldbik W M, Vredenberg A M, Habraken F H P M. Structure of sputtered silicon suboxide single- and multi-layers. Thin Solid Films 420-421: 382-385. 2002 24 Mandelbrot B B. The Fractal Geometry of Nature. San Francisco, Freeman. 1982 25 Wong P, Howar J. Surface roughness and the fractal nature of rocks. Phys. Rev. Letters 57: 637-640. 1986 26 Katz A J, Thompson A H. Fractal sandstone pores: Implications for conductivity and pore formation. Physics Review Letters 54: 1325-1328. 1985 27 Sinha I, Mansdal R K. Simulation studies on the nature of fractal dimensions of glass-ceramics at percolation threshold. Journal of Materials Science 38: 3469-3472. 2004 28 Bonn D, Kellay H, Prochnow M, Ben-Djemiaa K, Meunier J. Delayed Fracture of an Inhomogeneous Soft Solid. Science 280: 265-267. 1998 29 Lee, S. K. & Stebbins, J. F. Extent of intermixing among framework units in silicate glasses and melts. Geochim. Cosmochim. Acta, 66, 303-309. 2002 30. Mysen B. Physics and chemistry of silicate glasses and melts. Eur J Mineral 15: 781-802. 2003 31. Mysen B O, Virgo D, F. A. Seifert. Relationship between properties and structure of aluminosilicate melts. American Mineralogist, 77: 88-10. 1985
© Copyright 2026 Paperzz