Andrew Richard Lang

On Topographically Identifiable Sources of Cathodoluminescence in Natural Diamonds

An extensive examination of the cathodoluminescent emissions from natural diamonds has been performed with due regard to their inhomogeneity, correlating cathodoluminescence properties point-by-point with the local crystal lattice texture and imperfection content as revealed by other topographic techniques (in particular X-ray topography). Some dozens of crystals have been examined, mainly prepared in the form of cut and polished sections but in some cases as whole stones in their natural state. The cathodoluminescence observations have been made by visual microscopy, by photomicrography, and by spectrum topography’ with spatial resolution down to 5 pm. Particular attention was devoted to those crystals, not uncommon, whose growth stratigraphy included 2ones of type II (ultraviolet-transmitting) diamond intercalated within regions of the more usual type la (ultraviolet-absorbing) diamond. These type II zones prove to be particularly rich in fine structure within their patterns of cathodoluminescent emission, and in the spectral variety of their emissions. Joint cathodoluminescence topographic and X-ray topographic examinations were made on all specimens. Where feasible, the specimens were also characterized by ultraviolet transmission topographs, and by topographic recording of the anomalous spike diffuse X-ray reflexions. Many cathodoluminescence emissions (including both well-known and littleknown spectral systems) were discovered to have clearly defined topographically localized sources, e.g. dislocation lines or regions which had sustained natural a-particle irradiation. Some findings among many of this nature which are set out in detail concern the emission system (known as H3) which has zero phonon line at 2.46 eV and strong coupling to phonons of 40 meV energy. Its sources include curvilinear growth bands in regions where crystal growth has been of non-faceted, ‘cuboid’ habit rather than of the usual {111} faceted habit, slip traces and individual dislocation lines in matrices of type II character, and occasionally, in similar matrices, (lOO)-orientation platelets ranging from ca. 1 pm to several tens of micrometres in diameter. The H3 system emission from the platelets is more than 90 % linearly polarized with E vector in the platelet plane. (These platelets also emit in the near infrared, at energies of ca. 1.25 eV.) Another emission system with zero phonon line at 2.46 eV, but with only weak phonon coupling (dominant phonon energy ca. 66 meV), was found solely in emissions from the natural radiation-damaged rinds of diamonds, or from patches of natural radiation damage on their external surfaces. Noteworthy is the occurrence of dislocations blue-emitting and of dislocations emitting the H3 system in close juxtaposition within type II matrices. The deep blue broad-band spectral emission from dislocations is strongly polarized with E vector parallel to the dislocation line. The H3 system emission from dislocations is unpolarized. In dislocation-rich type II crystals possessing a mosaic texture the blue emission from dislocations is the dominant source of visible cathodoluminescence at room temperature. Evidence bearing upon the relation of the visible {100} platelets to the submicrometre size {100} platelets which give rise to the anomalous ‘spike’ diffuse X-ray reflexions is examined: as far as their X-ray diffracting properties show, they are indistinguishable.

Dissociation of dislocations in diamond

Weak-beam electron microscopy has been applied to study the dissociation of dislocations in a type II a diamond. Dislocations with Burgers vector ½[11̅0] on (111) glide planes have been found to be dissociated into two Shockley partials, with separations between 2.5 and 4.3 nm, and extended nodes and dissociated dipoles have also been observed. The stacking fault energy has been determined from the mean and distribution of the separations of the partials to be 279 ± 41 mJ m-2. The behaviour of dislocations in diamond appears to be similar to that of dislocations in Ge and Si.

On the dilatation of synthetic type Ib diamond by substitutional nitrogen impurity

Sequences of high Bragg-angle (0B = 74°) double-crystal X-ray topographs taken at the SRS (Daresbury, U.K.) have yielded precise measurements of lattice parameter differences between growth sectors of different crystallographic forms in a large undoped synthetic diamond whose type Ib infrared absorption spectrum (principal peak at 1130 cm-1) indicated atomically dispersed nitrogen, singly substituting for carbon, as the only detectable impurity. The plate-shaped specimen, polished parallel to (110), 5.0 x 3.2 mm 2 in area, 0.7 mm thick, possessed an unusually well developed (110) growth sector containing nitrogen impurity concentration of only ca. 10-6, which served as an internal standard of pure-diamond lattice parameter with which lattice parameters of nitrogen-containing growth sectors were compared. The specimen’s suitability for precision diffractometry was checked by comprehensive tests using optical microscope techniques, cathodoluminescence and single-crystal X-ray topography. The double-crystal combination was silicon reference crystal, asymmetric 175 reflection, with diamond specimen symmetrical 440 reflection. The principal measurement was the increase of the lattice parameter, a0, of the (111) growth sector (nitrogen content 88 + 7 parts per 106 atomic) relative to that of the (110) sector: Aa0/ a0 = 1.18 + 0.07 x 10-5. In terms of measured infrared absorption coefficient at 1130 cm-1, this gives Aa0/a0 = (2.95 + 0.27) x 10-6 [p(1130 cm-1)/cm-1], which is believed to hold for growth sectors of all crystallographic forms. Combination with the nitrogen assay findings of Woods, van Wyk & Collins (Phil. Mag. B 62. 589-595 (1990)) provides a direct relation to cN, the fractional atomic concentration of substitutional nitrogen, as Aa0/ a0 = (0.14 + 0.02)cN, which indicates that the effective volume of a single substitutional nitrogen atom in diamond is 1.41 +0.06 times that of the carbon atom it replaces. This substantial dilatation conflicts with several models for the substitutional nitrogen structure.

X-ray Topographic Studies of Dislocations in Iron-Silicon Alloy Single Crystals

Thin plates cut from melt-grown single crystals of iron +3.5% silicon were examined by an X-ray transmission topographic method and the spatial arrangement of individual dislocations and of dislocation arrays was investigated. The directions (including the sense in some cases) of dislocation Burgers vectors were identified. It was confirmed that Burgers vectors lie along <111>. Reactions such as ½[1̄11] + ½[111̄] = [010] were not observed among the individually resolved dislocations. The minimum separation of dislocations for easy individual resolution was 3 μm with CoKα radiation and 5 μm with AgKα. It was demonstrated that a one-to-one correspondence exists between dislocation outcrops and the etch pits produced by the dislocation-etching technique of Šesták (1959). The experiments showed that transmission X-ray topography when applied to metals of moderately heavy atomic weight can give a clear picture of the dislocation configuration in specimens sufficiently thick to be fully representative of the bulk material.

On voidites: a high-resolution transmission electron microscopic study of faceted void-like defects in natural diamonds

In natural diamonds of optical classification type la , nitrogen is the major identified impurity and is distributed mainly in point defects known as A defects (probably a pair of nitrogen atoms substituting for a pair of adjacent carbon atoms) and B defects (probably four substituted nitrogen atoms tetrahedrally surrounding a carbon vacancy), and also in the electron-microscopically visible platelet precipitates on {100}. This paper is concerned with other electron-microscopically detectable defects, discovered by R. F. Stephenson (Ph.D. thesis, University of Reading (1977)), that lie in {100} planes in circumstances strongly suggesting that they result from the decomposition of platelets. High-resolution electron microscopy shows these defects to be {111}-faceted cavities. They behave as pure phase-contrast objects whose interior density does not exceed about one-third that of the diamond matrix: we call them ‘voidites’. The experimental background to voidite observation is reviewed, including electron-microscopic measurements on normal {100} platelets and models of their structure, and the optical, X-ray diffraction and cathodoluminescence evidence for unusually large platelets whose presence, together with a relative richness in B defects, indicates an environment in which voidites are likely to be discovered. Almost all observed voidites are confined to sheets strictly parallel to {100}. Some voidite sheets occur in ‘ partial platelets’, where they replace part of the original area of normal platelet. Other voidite sheets occur within dislocation loops whose size and shape are similar to those of the peripheries of normal platelets in the specimen. Voidites occur in a wide range of sizes. The largest equiaxed voidites observed measure about 10 nm between opposite {111} facets, and the smallest resolved about 0.5 nm. Many voidites are elongated in one of the <110> directions in the plane of the voidite sheet: the most highly elongated voidites seen approach 100 nm in length, with diameters of a few nanometres. Variations in size, shape and number density of voidites, together with many other characteristics relevant to the microscopic processes of voidite formation, are discussed in detailed descriptions of about 40 voidite sheets occurring in partial platelets and dislocation loops in two diamond specimens. One specimen was free from both grown-in dislocations and dislocations associated with plastic deformation. It contained zones of highly elongated platelets and it appeared that transformation of a platelet into a voidite sheet surrounded by a dislocation loop was triggered by the mutual very close approach of platelets. The second voidite-containing specimen had suffered plastic deformation at some stage in its history, but did not exhibit direct evidence that glide dislocations had triggered the transformation. The Burgers vectors of 24 dislocation loops enclosing voidite sheets in the second specimen were determined. Twelve were of normal ½<110> type having a component ½a0 normal to the voidite sheet, and twelve were non-primitive, the Burgers vector being a0 normal to the voidite sheet (a0 is the diamond face-centred cubic (fcc) unit cell edge). The volumes of over 2000 individual voidites, representing all or major parts of 12 voidite sheets, have been measured. Values found for the ratio ∑V/Aa0 (where ∑V is the aggregate voidite volume in a sheet area A) averaged about unity for 9 sheets of generally similar, voidite-rich appearance. Other sheets are poorer in voidites of measurable dimensions: the ratios for two such sheets averaged 0.25. In the concluding analysis, a reaction involving A and B point defects is proposed for the production of platelets. Other reactions including voidites (but no dislocations) are suggested in which both platelet production and elimination might occur. For the dominating reaction, when a platelet is replaced by a voidite sheet surrounded by an interstitial dislocation loop, models are developed for the cases when the Burgers vector component perpendicular to the loop is either a0 or ½a0, with the assumption that the platelet nitrogen is dispersed partly into B defects and partly into the voidites. The predicted values of ∑V/Aa0 come out as about unity and as 0.25 (or lower) for the larger and smaller Burgers vectors, respectively.

Dislocations in diamond and the origin of trigons

Combined X-ray topographic and phase-contrast micrographic studies of natural diamond octahedral faces show that pyramidal trigons are always associated with dislocation outcrops. No dislocations outcrop at flat-bottomed trigons. On the other hand, absence of trigons does not necessarily imply absence of outcropping dislocations. The X-ray evidence strongly supports the etch-pit hypothesis for the origin of trigons. A dependence of trigon pit slope upon the Burgers vector orientation of its associated dislocation was not conclusively demonstrated.

On the Measurement of Population Density and Size of Platelets in Type Ia Diamond and Its Implications for Platelet Structure Models

This paper is concerned with the way that nitrogen impurity in type Ia diamond is distributed between the defects responsible for the principal impurity-dependent infrared absorption maxima in the 7-11 µm wavelength range. The defects involved are the platelets on diamond {100} planes, which produce the B´ absorption peak at 7.3 µm, and the A and B defects identified by their absorption maxima at 7.8 µm and 8.5 µm, respectively. The relation between the strength of the B´ absorption and the platelet area per unit volume measured electron-microscopically is found to be Ap/µm-1 = (9.0 ± 2.1) x 10-3I(B´)/cm-2, where Apis the platelet area per unit volume, expressed in square micrometres per cubic micrometre, and I(B´) is the integrated absorption, i. e. the area under the B´ peak, expressed in reciprocal centimetres squared. Features of the experimental methods employed include application of weak-beam microscopy for accurate measurement of platelet size and shape, counting a large sample platelet population in specimen thicknesses accurately determined in terms of extinction distance and measurement of infrared absorption through the same thin slice in which at a closely adjacent point platelet counting was performed. The constraints that the relation between Ap and I(B´) imposes on platelet structure models, when taken in conjunction with the correlations between A, B and B´ absorptions established by G. S. Woods (Proc. R. Soc. Lond. A 407, 219-238 (1986)), are analysed. Currently accepted models of nitrogen-containing A and B defects are assumed, and the analysis presupposes that platelets are formed by the conversion of A defects into platelets and B defects. It is found that conversion of A defects cannot supply sufficient nitrogen to form platelets containing four nitrogen atoms per area unit a20(a0 is the diamond face-centred cubic unit cell edge), but could form platelets with two nitrogen atoms per unit area a20;. Another constraint on platelet structure appears out of the analysis when considering the known expansion of the platelet cell by ca. 0.36 a0 normal to the platelet plane. It is that the atomic packing density in the platelet structure cannot be more than about three-quarters that of the perfect diamond structure, and this restriction applies whether the platelet is nitrogen-free or contains all the nitrogen released from A defects.

X-Ray Bragg Reflexion, 'Spike' Reflexion and Ultra-Violet Absorption Topography of Diamonds

The relation between X-ray diffracting properties and ultra-violet absorption has been studied point by point on polished plates of diamond having thicknesses from 0-1 mm to more than 1 mm and edge lengths from 2 mm to more than 1 cm. Most specimens exhibit marked non-uniformity of ultra-violet absorption. The intensity of the ‘spike’ X-ray reflexions was found to be roughly proportional to ultra-violet absorption, point by point, a t least in the range of weak and moderate spike intensities. Hence it is concluded that the platelets responsible for the ‘spikes’ are also responsible for the characteristic type I absorptions. However, high values of spike intensity and ultra-violet absorption have only been observed in relatively perfect crystal regions with a low density of grown-in dislocations as determined by X-ray topography; and it is suggested that in imperfect regions containing many grown-in dislocations impurity precipitation in the form of platelets coherent with the matrix may have been inhibited or pre-empted by precipitation in some other way. Thus the absence of ‘ spikes ’ and characteristic type I absorptions may not necessarily imply absence of impurity. On the other hand, some regions of high ultra-violet transparency can be highly perfect; hence lattice imperfection is not necessarily associated with type II optical characteristics. A comparison of ‘spike’ intensity in equivalent cube directions, made on a specimen selected for the high definition of its ‘spike’ topograph features, showed that where platelet precipitation was present it had occurred equally on all cube planes. The sharpness of boundaries between regions of high and low ultra-violet absorption and the association of such boundaries with growth stratifications revealed by Bragg reflexion topographs suggest that they correspond to variations of concentration of grown-in impurity in the crystal, no detectable subsequent diffusion having taken place.

Stacking fault contrast in X-ray diffraction: a high-resolution experimental study

Stacking fault fringe patterns and related dynamical diffraction effects have been studied by X-ray section topographs in the case of a fault about 1mm2 in area contained in an unusually perfect and almost dislocation-free type Ia natural diamond cut and polished into a plate 1 mm thick. The fault, a growth accident, lay on an octahedral plane inclined to the local octahedral growth surface and was bounded by a pair of partial dislocations parallel to <110>. Images of the fault were recorded with AgKα1, MoKα1 and CuKα1 radiations. Those obtained with AgKα1 and MoKα1 represented the low absorption case, and were compared with computer simulated images. In the case of moderately high absorption (CuKα1) both diffracted beam (Kg-beam) and direct beam (K0-beam) section topographs were recorded. The sign of contrast of the first fringe (i.e. of that fringe generated closest to the outcrop of the fault with the specimen surface) was observed in both K0-beam and Kg-beam patterns. These and other combinations of first-fringe contrast observations showed that the fault was intrinsic. Several departures from perfect-crystal diffraction behaviour were noted, and interpreted in terms of a long-range curvature of the lattice (which produced deviations in the trajectories of interbranch scattered rays) and short-range strains which reduced the strength of anomalous transmission of Bragg reflexions and superimposed a background intensity upon the dynamically-diffracted image. From microdensitometric scans across images of faulted and unfaulted regions of the specimen, estimates of the background were extracted, and amounted to about 1/3 the intensity of the ‘perfect-crystal’ pattern at the centre of the section topograph image in the case of AgKα1 and MoKα1 radiations.

A high-resolution section topograph technique applicable to synchrotron radiation sources

Asymmetric Bragg reflection from the surface of a plane crystal monochromator is employed to provide the spatially narrow beam (width less than 10 pm) required for high resolution in X-ray section topographs. Details are given of apparatus in which the monochromator is traversed back and forth to even out local irregularities in its reflecting power, and a precisely positionable screen is interposed between the monochromator and specimen to cut out radiation reflected from imperfections below the monochromator surface. The system was designed to be tolerant of some imperfection in the monochromator and this objective was achieved. With Cu K~I radiation, a natural diamond (111) face and the asymmetric 220 reflection, a lateral compression of the beam by a factor of 22.6 was obtained. Section topographs are exhibited in which the spatial width of the incident beam is only 7 pm. A benefit of this technique is the possibility of reducing the wavelength bandwidth effective in recording the topograph to less than that of the K~1 emission profile, and hence of reducing the contribution of dispersion to diffusion of the topograph image. The applicability of the method to synchrotron radiation sources is discussed.