Observations of Microcrystalline Plagioclase Spherulites with the Transmission Electron Microscope

Two thin sections of macroscopic plagiodase spherulites of approximately cm diameter found in a rhyolitic glass have been studied with the transmission electron microscope (TEM). Orientations of the thin sections were chosen to give views down and perpendicular to the major fiber axis. The crystalline fiber phase is high albite microtwinned on the (010) composition plane, elongated in the major growth direction, [001]. Fiber morphology is non-polygonal with an average fiber diameter of 2000 , perpendicular to c*. Fibers are separated by a non-crystalline residuum layer of approximately constant thickness (300-500 ,). Microtwinning relationships, as well as selected area diffraction (SAD) patterns, reveal both crystallographic and noncrystallographic branching with the former unexpectedly dominant.


INTRODUCTION
Current theories and definitions characterizing the nature of spherulitic crystal growth and resultant morphologies are based on the observation of relatively coarse grained spherulites (fiber diameters greater than 10-3 cm) visible under the light microscope.As an example, the definition ofa spherulite as, "... a radiating array of crystalline fibers all having the same fiber axis and possessing, therefore, the unusual property of branching in such a way that the crystallographic orientation of a branch departs slightly, but appreciably, from that of its parent ..." (Kieth and Padden, 1963) was obtained from observations of coarse grained, laboratory grown, organic polymer spherulites.These observations were then extrapolated to characterize non- resolvable, ultra-fine grained spherulites (fiber diameters to less than 10-5 cm).The question arises: Do current theories and definitions regarding spherulite morphology and growth mechanisms, obtained from observations of coarse grained spherulites, still apply to naturally occurring, ultra-fine grained spherulites when direct observation of these phenomena is possible?
In this study, extremely fine grained naturally occurring spherulites, found in an extrusive rhyolite glass from Modoc Butte, Modoc County, California, have been analyzed with a transmission electron microscope (TEM) which is capable of more than a three order of magnitude increase in resolution beyond light microscopes.With the added ability of obtaining selected area diffraction (SAD) patterns, which give an exact record of the crystallographic orientation of the subject in view, the TEM is a powerful tool in the analysis of spherulitic morphology and growth.

SAMPLE PREPARATION
Two thin sections were cut from macroscopic spherulites approximately 1 cm in diameter found in the sample from Modoc Butte.Sample 1 was prepared by cutting a spherulite near its edge to obtain a thin section giving a view down the major axis of the spherulitic crystal fibers.Sample 2 was prepared by cutting a thin section approximately through the center of a spherulite giving a view perpendicular to the radially oriented spherulitic fiber axes.After the thin sectioning described above, specific areas of interest were ion thinned and carbon coated for TEM observation.
Samples were analyzed with the JEM-100C TEM which allows direct observation in bright field and dark field, observation of selected area diffraction (SAD) patterns, and semi-quantitative chemical analysis through energy dispersive X-ray (EDX) analysis; all of which were used.

OBSERVATIONS Sample 1
Sample 1 micrographs (1, 2 and 3), taken in bright field, give a view down the major spherulitic fiber axis.SAD patterns of these micrographs reveal an a* b* orientation of high albite twinned on the (010) composition plane; EDX analysis confirms a sodic plagioclase as the crystallized phase.Fiber morphology ranges from roughly polygonal in section (1) to highly chaotic (2).Fiber diameter averages 2000/ with each fiber separated from others by a non-crystalline substance.This substance, assumed to be a glass resi- duum, forms a layer of surprisingly constant thickness (300-500 ,).Un- fortunately, this glass residuum layer is unresolvable by EDX analysis on the microscope used.However, comparison of the EDX analysis ofthe crystal- line fibers of the spherulite with the EDX analysis of glass external to the spherulite leads to the conclusion that the residuum layer is enriched in Si, K and Ca and depleted in Na and A1.
Extremely fine microtwinning (to less than 100 A) is ubiquitously observed in the fibers.Collinearity of microtwinning between adjoining fibers is often seen indicating highly similar, if not identical, crystallographic orientation.
In (1), this occurs occasionally, in (2), all microtwinning is collinear.Un- IO00A FIGURE Photomicrographs (top) and 2 (bottom).Description in text.fortunately, SAD patterns of the micrographs above cover too large an area to resolve whether crystallographic continuity exists between these fibers.Notably, individual microtwins do not cross the glass residuum layer even though collinear microtwinning exists between neighboring fibers.Close examination of micrograph 1 reveals a possible point of nascent fiber branching to the middle right.Here one can see a "daughter" fiber which is encircled by "arms" of its parent.Note the highly constant crystallographic orientation shown by the microtwinning, but the discontinuity of discrete twins.
In micrograph 3, fibers are seen to exist in bunches exhibiting a general, ifnot exact, collinearity ofmicrotwinning, suggesting crystallographic branch- ing within each bunch.Fuzziness of the right bunch is due to the fiber axis being slightly tilted from parallel to the viewing direction.Also, note the convolute shape of the fiber to the lower left center which has at least three distinct crystallographically continuous branches.

Sample 2
Micrograph 4, taken in dark field, is a view of fibers with their major axis perpendicular to the viewing direction; the growth direction is up and to the left.The SAD pattern reveals a b* c* orientation of high albite twinned on the (010) composition plane.Fiber morphology exhibits irregularly thick- ening and thinning branches elongated in the growth direction, c*, with many smaller, highly irregular protrusions and intrusions mirrored in negative on neighboring fibers.Most of the fibers are truncated at some place along the major axis indicating intense competition during fiber growth.Unconnected branches in the upper part of this micrograph reveal that growth has occurred in directions other than the major axis.All branches exhibit collinearity of microtwinning which the SAD pattern confirms is crystallographic branching.
Note that individual microtwins are not continuous between all parts of the branches, only where the fibers are connected in a straight line along c*.Also, note the general constant thickness of the glass residuum layer.

SUMMARY OF RESULTS
A general summation of the characteristics of naturally occurring, ultra-fine grained plagioclase spherulites can be drawn from the observations above.These characteristics are: --Non-polygonal external character of fibers.
--.Fibers grow in directions other than the major axis.
--Crystallographic and non-crystallographic branching occur with domin- ance of the former, causing formation of crystallographically continuous fiber bunches.
Unfortunately, interpretation of the above data is in several places am- biguous.The two dimensional sections necessary in microscopical obser- vations result in an incomplete picture of the three dimensional morphology they represent and only hint at the possibility that the fibers circled and turned as they grew.Also, the distinction between primary and secondary crystallization is difficult.It is assumed that the larger branches are primary growth features and the minor protrusions and intrusions secondary, as suggested by the literature (Kieth and Padden, 1962).However, the possibility exists that all branches are primary features and each perturbation is either a suppressed site of major branch formation or the base of a major branch growing out of the plane of the section.In addition, the presence of considerable asterism in many SAD patterns makes determination of crystallographic branching difficult; here it is assumed that collinear micro- twinning indicates crystallographic branching.Even though these problems exist, the generalized characteristics ofmicrocrystalline plagioclase spherulites listed above can be made with confidence.DISCUSSION Lofgren (1971, 1974), has shown that for natural silicate melts, there exists a progression of crystal forms for plagioclase which is dependent upon the degree of supercooling to which the melt is subjected.This progression, in order of increasingly supercooled melt, is as follows: (1) equant, (2) skeletal, (3) dendritic, (4) spherulitic.Also, he stated that there must be all gradations between end members.
In the search for naturally occurring spherulites to study, attempts were made to obtain those which would be as close to end member form as possible.They are extremely microcrystalline with discrete fibers totally unresolvable under a light microscope and have a perfect spheroidal shape.Recalling the definition of a spherulite given above where, "... the crystallographic orien- tation of a branch departs slightly, but appreciably, from that of its parent and comparing it with the observations of microcrystalline spherulites shown above in which crystallographic branching is seen to occur as the dominant branch form, we see that the definition incompletely describes these spherulites.
Interestingly, the theory describing spherulitic (non-crystallographic)   growth mechanisms (Kieth and Padden, 1964a and b), is virtually identical to that used to explain dendritic (crystallographic) growth mechanisms (Chalmers, 1964 and Maasen, 1978), whereby the cellulation of a growing nucleus occurs because of the presence of a supercooled, impure, viscous host melt.Both theories predict the occurrence of a residuum layer of constant thickness surrounding crystalline fibers having primary and secondary branches, with the only difference being whether the degree of supercooling allows non-crystallographic or crystallographic branching to occur, neither of which was exclusively seen here.Notably, the most similar descriptions within the literature to the crystal growth forms seen here, were made by a dendrologist, Saratovkin (1959), who observed many unusual crystal- lographic, as well as non-crystallographic, growth forms, though he never correlated them with spherulites.
The degree to which crystallographic branching occurs in natural plagio- clase spherulites, has not been recognized before because the resolution limits of light microscopes (approximately l0 -4 cm) precluded its observation.To form a "perfectly" rounded spherulite exhibiting "exclusively" non- crystallographic branch characteristics as seen under a light microscope, branches averaging 2000 , in diameter could be in bunches 10 fibers across with approximately 75 fibers per bunch.Allowing for the unknown effects of fiber competition, i.e. truncation, far less than half of the fiber branches need be non-crystallographically oriented to give rise to an "ideal" light microscope observed spherulite.In this study, bunches over 6 fiber diameters in width were recorded.

CONCLUSION
In all, it can be concluded that current theories and definitions regarding spherulitic growth mechanisms and the resultant morphologies are essentially correct in predicting many of the phenomena observed in the naturally occurring spherulites described above, i.e. fibers existing in a generally radially oriented pattern, all having the same crystallographic axis, separated by a glass residuum layer of approximately constant thickness.However, the presence of fiber bunches with crystallographically constant orientation within each bunch indicates that crystallographic branching, though pri- marily restricted to growth away from the spherulite center because of fiber competition for pregnant melt, is the major form of branching, not non- crystallographic branching as previously hypothesized.Therefore, it is here indicated that these naturally occurring spherulites, and quite possibly natural spherulites in general, are actually an intermediate crystalline growth form be- tween end member spherulites and dendrites that can not be considered equivalent with coarse grained, laboratory grown spherulites.Also, the complexity of the fiber growth patterns puts these spherulites into a different class from the perfect spicule-star morphology as shown by Kieth and Padden (1964a).Further work with the TEM, however, is definitely needed to more quanti- tatively describe spherulites in volcanic rocks and to investigate vitreous rocks in much more detail.