The following article is a component of the July 1999 (vol. 51, no. 7) JOM and is presented as JOM-e. Such articles appear exclusively on the web and do not have print equivalents.

Solidification: Research Summary

Monitoring the Solidification of Single-Crystal Castings Using High-Energy X-Ray Diffraction

D.W. Fitting, W.P. Dubé, and T.A. Siewert
JOM-e Logo

A noninvasive x-ray technique was developed at the U.S. National Institute of Standards and Technology to monitor the solidification of single-crystal castings. X-ray energies of 150 keV to 320 keV have sufficient energy to perform transmission x-ray diffraction on a 17 mm thick nickel-alloy specimen. Laue diffraction images were obtained from the mold-encased casting even though the x-ray path (more than 1 m) through a directional-solidification furnace included a variety of intervening furnace components. The x-ray method was capable of sensing changes in the physical state of the casting (liquid or solid) and measuring the fraction of solid in the region of dendritic solidification.


Analytical x-ray diffraction is conducted in the reflection geometry using x-ray energies of 5–17 keV, returning crystal-structure information only from the surface and a few tens of micrometers beneath it. Since the penetration depth of low-energy x-rays is shallow, traditional x-ray diffraction (XRD) is, thus, unable to probe the interior of thicker structures. Synchrotron radiation, producing x-ray energies of 100 keV and higher, permits the study of specimens up to a few millimeters thick.1,2 In the study reported here, it was shown that by using higher x-ray energies (100–320 keV) and a transmission configuration, melting and solidification of a thick metal specimen within a mold can be noninvasively discerned.

Using XRD to study metal solidification and phase changes is not new.3–10 However, most of this research used very thin specimens (a few millimeters at most), furnaces with low attenuation x-ray windows (beryllium, graphite, or polyimide), and low x-ray energies (50 keV or lower). Work by Green11 extended XRD investigations to energies exceeding 150 keV. His flash XRD system also provided the facility for studying structural changes during dynamic events, such as crystal growth. Others, including Bechtoldt et al.,12 Kopinek et al.,13 Black et al.,14 and Reimers et al.,15 have employed high-energy XRD for studies of stress and texture in thick (up to 12.7 mm) steel specimens. For more detail on attentuation factors and XRD, refer to the sidebar Attentuation and XRD: A Primer.

At the National Institute of Standards and Technology (NIST), a technique based on transmission XRD was developed to study the solidification of a single-crystal turbine blade casting within its mold. High-energy x-rays (150–320 keV) penetrate through material surrounding the casting and produce a distinctive diffraction pattern that clearly indicates whether the sampled region is liquid or solid. A real-time transmission Laue x-ray image of the casting shows an ordered pattern of x-ray scattering (diffraction spots) from the solid and a diffuse ring of scattering from the liquid. The dramatically different spatial pattern provides a high-contrast, unequivocal spatial discrimination of the physical state of the alloy.

Attentuation Coefficients

Interactions of a narrow beam of monoenergetic x-rays with matter may be characterized with a total mass attenuation coefficient (µ/). The intensity of x-rays transmitted (at perpendicular incidence) through a slab of material is given by


where I is the transmitted x-ray intensity, I0 is the incident x-ray intensity, µ/ is the mass attenuation coefficient (a function of x-ray energy), is the material density, and x is the slab thickness.

A linear attenuation coefficient (µ), defined as the product of the mass attenuation coefficient and the density, is useful for comparing the relative transmission through different materials of the same thickness. The reciprocal of the linear attenuation coefficient (material thickness that reduces the intensity to 1/e) is often used as a measure of the effective penetration depth for x-rays. Plots of 1/µ for several materials are shown in Figure A. Raising the x-ray energy increases transmission, because the attenuation coefficient decreases with energy.

The total mass attenuation coefficient may be expressed in terms of the sum of partial mass attenuation coefficients,16 which reveals the contributions by the photoelectric (PE), Compton (Comp), and coherent (coh) processes.


Plots of attenuation coefficients for nickel as a function of energy are shown in Figure B. The dominant interactions of x-rays with matter are the photoelectric process at low energies and Compton scattering at higher energies. Although coherent scattering (which gives rise to diffraction) is not a large contributor to the total attenuation, this interaction does occur at high energies.

X-Ray Diffraction

Figure A Figure B Figure C
Figure A. The reciprocal of the linear attenuation coefficient, µ, as a function of x-ray energy. The physical meaning of 1/µ is the effective penetration depth. Figure B. Partial and total linear attenuation coefficients for nickel as a function of x-ray energy. Figure C. The atomic scattering factor at particular x-ray energies for nickel, plotted as a function of scattering angle (2).

The XRD intensity of an unpolarized x-ray beam from a small crystal when the Bragg condition is satisfied may be calculated from the kinematical theory of diffraction.17 Although the absolute intensity is not measurable, the total diffracted energy (integrated intensity) can be measured. The total diffracted energy is given by17


where E is the total diffracted energy; I0 is the incident x-ray intensity; is an angular rate of rotation of the specimen, required to insure that all area under the interference functions has been integrated; e is the charge of an electron; R is the distance from the electron in the crystal where the intensity is measured; m is the mass of an electron; c is the speed of light in a vacuum; 2 is the angle between the incident and scattered directions; 0 is the dielectric constant of free space; F is the structure factor of the crystal; e–2M is the Debye temperature factor, and 2M = 16<us>2 (sin2/2), where <us> is the component of the mean displacement due to temperature effects in the direction normal to the diffracting planes; V is the volume of a small segment of the crystal illuminated by the incident x-ray beam; va is the volume of the unit cell of the crystal; and is the x-ray wavelength.

The energy dependence of diffraction comes into Equation C through the energy-dependent atomic scattering factors (f),18,19 which are included in the structure-factor term FF*. The Lorentz-polarization factor term is also a function of x-ray energy, through the scattering angle 2. In the forward direction (2 = 0), scattering from all electrons of an atom is in phase. The atomic-scattering factor is a function of sin/; the value of f decreases for off-axis scattering and is dependent of the wavelength of the scattered x-rays. The energy and angular dependence of the scattering factor is demonstrated in Figure C, which plots the atomic-scattering factor for nickel. For high energies, the scattering becomes increasingly forward-directed.

Often, a simple form of the diffraction conditions is used. The Bragg equation

where n is an integer, is the wavelength of the x-rays (nm), E is the energy of the x-rays (keV), d is the lattice spacing (nm), and 2 is the angle between the incident and diffracted x-ray, expresses these conditions. The diffraction interaction is, thus, a spatial () and energy (E) filter. For a given set of lattice planes, x-rays of higher energy coherently interfere at smaller angles. An x-ray imager placed in the diffraction field will record the spatial location of diffracted x-rays. An energy-sensitive detector placed at the same location as the imager would record energy peaks associated with each of the diffraction spots.

Implications for High-Energy Transmission Diffraction

Probing the interior of a casting requires a transmission configuration. X-ray energies of more than 100 keV are needed to penetrate the refractory oxide mold (5–10 mm wall thickness) and casting specimen (1–20 mm thick).

Consider a beam of x-rays incident on a crystalline specimen contained within a casting mold (Figure D). The primary x-ray beam is attenuated as it passes through the mold wall and a portion of the specimen to a location where coherent scattering occurs. The scattered x-ray is attenuated along its exit path through the remaining specimen and exit mold wall. Attenuation losses along entrance and exit paths are minimized by raising the energy substantially above (5–20 keV) that used in conventional x-ray diffraction systems to 150–320 keV.

Figure D
Figure D. The geometry used in developing an analytical model for transmission XRD by a casting specimen encased by mold wall.
The integrated intensity from a crystalline solid, given in Equation C, is the product of the squared structure factor (FF*) of the crystal, the Thomson-scattering amplitude from an isolated electron, and the irradiated volume of unit cells. F is a function of the types of atoms in the crystal (through the atomic scattering factor, f), the configuration of the unit cell of the crystal, and the particular lattice planes that are involved in the scattering. For a face-centered cubic (f.c.c.) crystal, such as nickel, the structure factor is equal to 4 f if the Miller indices (hkl) are unmixed (all odd or all even), but is zero if the indices are mixed. The structure factor squared gives 16 f2 for an f.c.c. material with lattice planes defined by unmixed Miller indices.

The dependence of the atomic-scattering factor for nickel was plotted in Figure C as a function of the scattering angle for several x-ray energies. At low x-ray energies, the atomic-scattering factor is quite large, even for large scattering angles. However, with high x-ray energies, f is significant only for small scattering angles (i.e., for a transmission geometry). In summary, low-energy (10–40 keV) x-rays are diffracted over wide angles, while high-energy (100–300 keV) x-rays diffract at narrow angles about the primary beam. For example, the scattering factor for nickel is 16.5 for 150 keV x-rays scattered at 3.5° from the direction of the incident x-ray beam (typical of our experimental geometry). The structure factor squared for this example is 16 x (16.5)2 or 4,356.

Because the irradiated volume (1 mm x 1 mm x 0.001 mm) contains many unit cells (for a nickel crystal va = 4.376 nm3), the ratio V/va is very large (2 x 1016). The large atomic-scattering factor in the forward direction and the enormous number of unit cells along the path of the primary beam (all with the same crystalline structure and orientation), in addition to the substantial intensity of high-energy x-rays that penetrate through a mold and specimen, account for the efficiency of high-energy transmission diffraction.


We have developed a model for transmission XRD that includes attenuation in the entrance mold wall, attenuation in a portion (x in sidebar Figure D) of the specimen to the point where diffraction occurs, the efficiency of diffraction (sidebar Equation C), and attenuation of the diffracted x-rays through the remaining portion of the specimen and the exit mold wall.

Figure 1a shows plots of the model-predicted efficiency of diffraction for a nickel specimen. The efficiency of transmission XRD is a strong function of both the specimen thickness and the incident x-ray energy. The broad peaks in the plots indicate an optimal range (100–275 keV) of x-ray energies for transmission XRD on these nickel specimens. The optimum energy yields the highest intensity of diffracted x-rays, which are able to penetrate through the specimen.

Figure 1a Figure 1b
a b
Figure 1. Plots of model-generated XRD efficiency from (a) a nickel specimen at thicknesses of 2–25 mm and (b) a nickel specimen 10 mm thick surrounded by mold material.
Figure 1b shows plots of diffraction efficiency for a nickel specimen 10 mm thick sandwiched between two pieces of mold material, each of the thickness indicated in the legend. Attenuation coefficients for the mold material (a mixture of alumina, silica, and zirconia) were computed using the XCOM software.20 The density of a sample of the mold material was 2.6 x 103 kg/m3. For x-ray energies over 150 keV, even mold-wall thicknesses as large as 7.5 mm decrease the XRD intensity by only about one-half.

Anticipating that the XRD technique would be used to probe the "mushy" zone (area of dendritic solidification containing both solid and molten material) of a casting, a specimen volume containing both liquid and solid was modeled. The liquid portion of the specimen was assumed to attenuate the incident and diffracted x-ray beams. The correct (lower) density of the liquid was used. The solid portion of the specimen attenuated both the incident and scattered x-rays as well and produced coherent scatter. A 6 mm thick section of N5 nickel-alloy casting was modeled, with various fractions of the thickness being composed of molten metal and crystalline solid. Energies from 20–500 keV were considered. The relationship between XRD efficiency and fraction of solid departed only slightly from a linear one. The curve had a slightly concave shape due to differences between the densities of the liquid and solid.

Figure E illustrates the key elements in the apparatus used for transmission XRD measurements. The x-ray tube produces a high-intensity, polychromatic source of x-rays restricted by collimators to a small circular beam. A fraction of the x-rays interacting with the sample is diffracted. An x-ray imager detects the x-ray pattern incident on its two-dimensionally sensitive area; an energy-sensitive detector may be used in place of the imager to measure the spectra of the primary and diffracted x-rays.

Figure E
Figure E. Components of the real-time, high-energy transmission diffraction apparatus.
Two sources of x-rays were used in the experiments. Both were metal-ceramic, tungsten-anode tubes with a beryllium exit window, intended for use in industrial radiographic imaging. A 160 kV system was used in early experiments; however, modeling indicated the need for x-rays of higher energy when a turbine-blade casting was to be probed. An increased x-ray tube voltage produces higher intensity characteristic radiation lines and a bremstrahlung spectrum and adds x-rays of higher energy that more easily penetrate the mold-encased casting. A new x-ray tube and high-tension transformer permitted experiments to be performed with tube potentials as high as 320 kV.

The choice of x-ray source diameter (focal spot size) of 1.2 mm or 4 mm (320 kV tubehead), or 0.2 mm or 3 mm (160 kV tubehead) was selected by activating one of the two filaments in the tube. The small (1.2 mm) focal spot size of the 320 kV tubehead proved most useful because its size corresponds to the diameter (1 mm) of the collimating apertures.

Controls on the constant-potential x-ray generator permitted x-ray tube voltages to be varied between 15 kV and 320 kV. A tube current of 3 mA was possible at 320 kV. For some experiments, a means for reducing the x-ray source intensity was required so the diffraction signals did not overwhelm the detector system. A lower intensity was obtained by modifying the controller. Tube currents as low as 0.05 mA with incremental changes of 0.05 mA were possible.

A beam-restricting collimator for the x-ray source was designed and fabricated. A triple-aperture design was employed to minimize beam divergence and optimize x-ray intensity.21 Interchangeable sets of 9 mm thick lead disks could be used to select beam diameters of 0.3 mm, 1 mm, 1.5 mm, or 2 mm. An alignment laser, placed in an opening in the base of the collimator, projected a visible beam along the same path as that of the x-ray beam. This permitted fast, safe, and simple alignment of the beam-stop and specimen.

The primary beam transmitted through the specimen is often of high intensity. Blocking this beam from the detector improves the contrast of images and prevents an energy-sensitive detector from being overwhelmed by high count rates. A small tungsten rod (4 mm diameter and 6 mm thick) was positioned in the center of the primary x-ray beam emerging from the specimen to act as a beamstop. X-rays diffracted from the sample passed to the sides of the beamstop and were imaged, while the undiffracted primary beam was severely attenuated. The tungsten disk was suspended by a thin graphite/epoxy composite support. The low attenuation in the composite minimized the shadow it cast on the x-ray images.

A real-time x-ray imager, designed specifically for XRD, was used for all experiments reported here. The scintillator screen was optically coupled to an image intensifier, which was then coupled to a charge-coupled device (CCD) camera sensitive to low-light levels. Fiber-optic coupling between the scintillator, image intensifier, and CCD camera transferred light very efficiently. The field of view of the imager was approximately 40 mm wide by 30 mm high. The imager was originally equipped with a 70 µm thick gadolinium-oxysulfide scintillator. The imager efficiency for high-energy x-rays was improved considerably by replacing the gadolinium-oxysulfide scintillator with a 6 mm thick fiber-optic glass scintillator. The fiber-optic scintillator produces an extremely low lateral spread of light and high-efficiency light transfer to the image intensifier. The glass scintillator also absorbed a much greater number of x-ray photons than the gadolinium oxysulfide, particularly at high energies.

Acquisition and storage of the video frames from the PAL-format imager was performed with an eight-bit, monochrome frame grabber installed in a personal computer. A coprocessor board, connected to the frame grabber, speeded image manipulations and frame averaging (to increase signal-to-noise ratio). Frame averaging could be performed at a rate of 25 frames per second. Image acquisition and processing software provided the ability to perform dynamic-range expansion and compression, image averaging, and image subtraction, as well as image filtering to enhance the acquired images.

A multiformat video cassette recorder was used to record radiographic and diffraction images during all experiments. The video tape served as a backup for direct image acquisition with the frame grabber and was useful in cases where the image was rapidly changing (e.g., dynamic events such as mold filling). Video frames could be digitized after the experiments by replaying the video tape through the frame grabber.


Our investigations to demonstrate the feasibility of performing transmission XRD on a specimen (contained in a mold) and to determine its physical state (solid or liquid) began with a gallium specimen, the properties of which permitted an independent confirmation of the physical state of the specimen.22 Gallium exhibits a substantial change (three percent) in density between the solid and liquid states, large enough to produce a discernible difference in brightness on a radiographic image. The ability to independently determine the position of the liquid-solid boundary (brighter versus darker areas of the radiograph) provided a means for validating the spatial performance of the XRD sensor.

Figure 2a Figure 2b Figure 2c
a b c
Figure 2d Figure 2e Figure 2f
d e f
Figure 2. Transmission XRD images probing the (a) solid, (b) solid/liquid boundary, and (c) liquid and radiographic images probing the (d) solid, (e) solid/liquid boundary, and (f) liquid. The upper portion of the radiographic image is brighter because the x-rays are less attenuated by the lower-density solid gallium.
A steady-state boundary was produced in an acrylic plastic test cell between the solid gallium (at the top because of its lower density) and the liquid gallium. The position of the probing x-ray beam could be moved into the solid or liquid. The XRD mode could be switched to a radiographic imaging mode by removing the collimating apertures from the x-ray beam. Figure 2 shows radiographic and XRD images with the x-ray beam positioned in the solid, on the solid-liquid boundary, and fully in the liquid gallium. The circular black area in the center of each radiographic image is the shadow cast by the tungsten beamstop. The x-ray beam is centered on the beamstop, so the dark area shows the position of the incident x-ray beam.


Figure 3 illustrates the key elements in the apparatus used for transmission XRD measurements during the melting and solidification of aluminum and copper. The x-ray tube, with supplemental lead shielding on the tube housing, is on the right. The collimated x-ray beam (1 mm in diameter) passes through the furnace (white cylinder in the center), through a quartz specimen tube, and into the metal specimen. A portion of the x-rays entering the specimen is diffracted. The x-ray imager (black cylindrical object on the left) detects and displays the XRD pattern. A tungsten beamstop is mounted between the furnace and the x-ray imager to intercept the primary x-ray beam.

A melting and recrystallization experiment was performed on an aluminum specimen in the gradient furnace. A 22 mm diameter, coarse-grained, polycrystalline, 99.999 percent pure aluminum rod in a quartz tube (2.2 mm wall thickness, 27 mm inner diameter), was placed in the furnace. Figure 4 shows the transmission Laue XRD patterns obtained during heating, melting, and resolidification of the aluminum.

As the rod was heated, the diffraction pattern changed, reflecting changes in the physical state of the specimen. The complex diffraction patterns in the first few frames of the image sequence were the result of interactions of x-rays with many crystals in the polycrystalline specimen. As the temperature was raised, the larger grains in the polycrystalline rod grew at the expense of the smaller ones (Ostwald ripening), and the diffraction pattern became simpler as the x-ray beam encountered fewer, but larger, crystals. Near the melting point (652°C), the diffraction pattern began to lose order. The Laue spot pattern degenerated into a diffuse ring of scattering when the aluminum was fully melted. The difference between the patterns produced by the solid aluminum and the liquid aluminum was dramatic and unmistakable. The Laue diffraction spots disappeared at the same time the diffuse ring formed. As the aluminum cooled below the melting point, the diffraction spots reappeared at the same time that the diffuse ring disappeared. The diffraction pattern for the resolidified aluminum specimen was simple, indicating an x-ray path encountering a few, large grains.

Figure 3
Figure 3. The apparatus for observing transmission diffraction during melting and solidification.
Figure 4
Figure 4. The sequence of transmission Laue diffraction patterns (160 kV, 1 mA) obtained as a polycrystalline aluminum rod was heated, melted, and then cooled.
In a similar experiment, changes were observed in the XRD pattern of copper during melting and solidification. As in the aluminum melting experiments, the solid copper produced diffraction images with many bright Laue diffraction spots. When the melting temperature of copper was exceeded, the ordered diffraction pattern disappeared and was replaced by a diffuse ring of x-ray scattering from the molten copper.


A resistively heated directional-solidification (DS) furnace (Figure 5) was used in the next set of experiments. The vacuum furnace was capable of producing 1,700°C temperatures. The x-ray source (A) and collimator (B) are on the right. The small-diameter x-ray beam passes through a borosilicate glass port 10 mm thick into the furnace. The beam then encounters molybdenum resistance-heater windings (D) (1.6 mm diameter). The wire is wound on an aluminum-oxide support tube (E) (100 mm internal diameter, 4.5 mm wall thickness). After passing through the coil support, the x-ray beam enters the casting mold (F) (6.4 mm wall thickness).

The cavity of the mold for the first melting experiment was 6 mm thick by 38 mm wide (producing a thin rectangular bar). The exit path for the x-rays, after diffraction from the specimen, was through 6.4 mm of mold, 4.5 mm of alumina, 1.6 mm of molybdenum, and 10 mm of glass. The real-time x-ray imager (G) was positioned outside the glass port to intercept the diffracted x-rays. The specimen was 780 mm from the x-ray source and 400 mm from the imager. The asymmetry in the location of the hot zone of the furnace within the bell jar was advantageous for placing the imager nearest the specimen. This shorter distance, from specimen to imager, yielded a larger angular field of view.

The x-ray source and imager were each attached to two-axis motion stages. The source and imager could be scanned in unison (horizontally and/or vertically) to probe different regions of the casting; however, by leaving the x-ray source fixed and scanning the imager, a large virtual field of view was achieved. The system was capable of moving the 130 kg x-ray tubehead in submillimeter increments and at speeds more than sufficient for following the liquid-solid boundary in a DS casting.

An early test, with no specimen in the furnace, disclosed that the primary x-ray beam intensity at the imager was not high enough to damage the imager. Therefore, no beamstop was required in these diffraction experiments. The large bright spot visible in many of the XRD images from furnace experiments is the primary x-ray beam.

Enclosures to provide radiation shielding were fabricated from 16 mm thick steel around the x-ray tube and 9.5 mm thick steel around the x-ray imager. Additional shielding was provided by lining the enclosures with lead sheet. Radiation surveys near the furnace indicated that the researchers could safely occupy areas near the furnace during operation of the XRD system.

Figure 5 Figure 6 Figure 7
Figure 5. The configuration of the DS furnace fitted with the XRD equipment. The dashed lines indicate lead-lined steel enclosures used for radiation shielding. Figure 6. A side view of an investment-casting mold used to produce single-crystal castings. Figure 7. A side view of the bell jar in the DS furnace after modifications. The tall foam spacer beneath the hot zone was added to enlarge the field of view for x-ray diffraction from the casting.

Figure 6 depicts a mold typical of that used in the single-crystal casting experiments. The mold is fabricated by repeatedly coating a wax replica of the part to be cast with a slurry of refractory oxides (alumina, silica, and zirconia). The slurry used for initial coats is very fine in order to produce a smooth inner coating. The slurry for outer coats is increasingly coarse; the outer surface of the molds had a root mean square (rms) roughness of about 2 mm. Roughness varied from mold to mold. Once the requisite mold thickness has been achieved, the wax replica is removed by heating the mold. The mold is then baked at high temperature to harden it. For our studies, we used molds with a circular, triangular, or rectangular cross section.

During casting experiments, a metal alloy charge is placed in the crucible of the mold. After a 1 Pa vacuum has been established, the gate valve is opened, and the ram raises the mold into the hot zone of the furnace. The alloy melts, filling the mold. Solidification of polycrystals initiates at the base of the mold in the starter block where the alloy contacts the water-cooled ram. The fastest-growing grain reaches the grain selector (a corkscrew-shaped section of the mold) first and blocks growth of all other crystals with differing orientation. If the thermal conditions are correct, growth of a single-crystal continues upward in the mold as the mold is slowly withdrawn (150 mm/h) vertically from the furnace. Polished and etched sections of the completed castings showed a dendritic structure with several phases and an absence of grain boundaries.

Initial experiments with the DS furnace indicated that the region of solidification in the casting was beneath the field of view of the XRD sensor. The hot zone of the furnace had to be turned off after melting occurred to drive the solidification upward into the region probed by the sensor. To correct this problem, we rearranged components in the furnace and added new elements.

The hot zone (alumina core with molybdenum windings) of the furnace sat directly on the chill plate. Solidification occurred at the location of the chill plate or below it. This area is inaccessible for XRD because of the locations of the diffusion pump inlet, the chill plate, and its cooling coils. A 60 mm thick piece of alumina foam was used as a spacer between the chill plate and the hot zone (Figure 7). The foam has a structural rigidity sufficient to support the hot zone, but a density (240 kg · m–3) low enough to provide a path of low x-ray attenuation, free of interfering structures. At the same time the foam spacer was added, the chill coils were raised slightly to preserve the high-temperature gradient in the furnace, which drives solidification in the vertical direction. The borosilicate glass ports on the bell jar were replaced by graphite-epoxy ports. Less of the incident and diffracted x-ray beams are attenuated in the ports, yielding a higher diffraction spot brightness. X-ray transmission improved by a factor of 1.5 compared to the configuration with the glass ports.

The alignment of the crystalline planes in the casting can vary rotationally about the axis of the casting and can deviate from the vertical axis. A means for orienting the probing x-ray beam to obtain strong diffraction spots was devised. The vertical ram of the furnace was altered to permit rotation of the mold (and specimen). The rotation allowed us to position the solidifying metal such that the Bragg condition was satisfied for particular lattice planes and diffraction spots could be observed. After the specimen-rotation capability was added to the furnace, strong diffraction patterns from the crystalline alloy castings could be obtained during every experiment.

A previously cast single crystal of N5 alloy 6 mm thick in a mold of 5 mm wall thickness was placed in the unheated DS furnace. Figure 8 shows the XRD image recorded for this experiment. A gray level profile through the primary beam and one of the diffraction spots is also shown. After the image was acquired, the x-ray imager was replaced with the energy-sensitive germanium detector. As mentioned earlier, XRD is manifested spatially as spots or as peaks in the transmission energy spectrum. A lead collimator (25 mm thick, 3.5 mm diameter hole) was affixed to the germanium detector to permit probing small areas of the XRD field.

Figure 8a Figure 8b Figure 9
a b
Figure 8. (a) A transmission XRD image (320 kV, 0.75 mA) of a mold-encased N5 specimen placed in the DS furnace. The bright area at the upper right is the primary x-ray beam. There are three XRD spots to the left and below the primary beam. (b) A gray level profile through the primary beam and the left diffraction spot. Figure 9. Transmission spectra with the collimated germanium detector positioned at the center of the primary beam or at the center of the diffraction spot left of the primary beam.

Figure 9 plots transmission spectra with the germanium detector positioned at the center of the primary beam or at the center of the diffraction spot to the left of the primary beam in Figure 8. The small spectral peaks in the 70–90 keV range are x-ray fluorescence lines from the lead collimators on the x-ray source and detector. The prominent spectral peak at 180 keV is produced by XRD in the nickel-alloy casting specimen. This peak is quite broad because the source and detector collimation were relatively coarse (i.e., there is modest angular divergence of both the source and diffracted beams).

XRD during Casting

A charge of N5 nickel-alloy was placed in the crucible at the top of the flat bar mold (6 mm thick by 38 mm wide cavity), and the mold was placed on the ram of the furnace. The load chamber and furnace were evacuated to a pressure of 1 Pa. The furnace temperature was ramped to 260°C over 30 minutes, ramped to 538°C over the next 30 minutes, and finally ramped to 1,093°C over 30 minutes. The gate valve separating the furnace was opened, and the mold raised slowly (500 mm/h) into the hot zone of the furnace. After a dwell time of 45 minutes, the furnace temperature was ramped to 1,566°C over a ten minute period. Molten N5 alloy began to fill the mold a few minutes after the end of the high-temperature ramp. The alloy-filled mold was allowed to soak at 1,566°C for 30 minutes before being withdrawn from the hot zone at a rate of 150 mm/h. A single-crystal structure formed above the grain selector as the casting solidified.

By removing the collimator of the x-ray source, a radiographic image of the mold could be formed, with an approximately 30 mm field of view. X-ray technique factors of 180 kV and 1 mA produced acceptable images of the casting within the hot zone of the furnace. Figure 10 shows the grain selector of the bar mold before any N5 had melted, the grain selector with a few drops of molten N5 at its top, and the grain selector immediately after it had been filled with molten N5 alloy. Filling the remainder of the mold with molten N5 was radiographically observed. After all of the N5 had melted, the furnace ram was used to slowly withdraw the mold through the temperature gradient established between the hot zone and the chill plate in the furnace.

The x-ray beam and imager were then positioned (using the remotely controlled motion stages) in the region of the alumina-foam spacer beneath the hot zone. We waited until the casting had withdrawn to the point where the x-ray beam was centered in the funnel-shaped area of the casting, just above the grain selector. The collimator was placed on the x-ray source to configure the system to observe XRD, and the tube potential and current were raised to 320 kV and 3 mA.

Figure 10a Figure 10b Figure 10c
a b c
Figure 10. Radiographic images of the grain selector of the bar mold (a) before, (b) during, and (c) after filling with molten N5 alloy.

The casting was manually rotated (by turning the ram) while observing the real-time transmission x-ray image. Before the solidification front advanced to the region probed by the x-ray beam, only the primary beam was visible. Eventually, solidification proceeded into the area probed by the x-ray beam and diffraction spots appeared. Rotating the specimen caused the spots to appear, move along hyperbolic paths through the center of the primary beam, and then disappear. The spots produced as different lattice planes in the single-crystal casting satisfied the conditions for Bragg reflection. Since a white beam source was used, there was not a singular angle where a spot appeared, but rather there was a small range of specimen orientations over which a spot was produced. The highest-intensity spot was obtained when the peak intensity in the bremstrahlung spectrum from the x-ray source had the correct wavelength for the particular Bragg reflection.

The single-crystal structure in a casting could be confirmed by spatially scanning the x-ray source and imager. If the diffraction pattern remained the same, the region scanned possessed a single-crystal structure. After confirming the crystal structure of a solidified portion of the casting, the x-ray source and imager were scanned vertically from the crystalline (solid) region through the region of dendritic solidification into a fully molten region of the casting. During the scan, the XRD spot intensity decreased.

The measured gray level of a transmission diffraction spot from a typical experiment where the x-ray beam was scanned vertically in the casting is plotted in Figure 11 along with a model (Lever) prediction of the fraction of solid versus temperature for N5 alloy. Since the temperature profile in the casting was not measurable in this experiment, we have adjusted the vertical and horizontal scale of the XRD data to coincide with the solidification model predictions. It is encouraging that the shape of the modeled and experimental curves are similar. The experimental data also appear to show the formation of the predicted second phase (break in the curve).

X-ray Topography of Dendrites during Solidification

In one of the casting experiments, the single crystal [001] axis was exactly aligned with the vertical axis of the furnace ram. With the x-ray beam slightly angled with respect to the (001) planes, a strong Bragg reflection was observed directly above the primary beam. The spot remained, but its mottled appearance changed as the specimen was rotated about a vertical axis. We think that this area (Figure 12) of the XRD image is a transmission topograph23 of dendrites in the solidifying casting. Bright areas in the spot represent reflections from crystalline solid with a [001] orientation. Dark areas are either molten alloy or misaligned areas of the solid.

A topograph is an image formed by the specular reflection of x-rays from crystalline planes; Figure 13 shows the geometry of topography. Reflections occur from lattice planes in correctly aligned dendrites. There are no reflections from the amorphous liquid surrounding the dendrites. As can be seen from the figure, there is a vertical exaggeration of the actual structure in the topograph. Additionally, there is a vertical enlargement due to divergence of the collimated x-ray beam. The horizontal enlargement of structure in the topograph is caused solely by divergence of the x-ray beam.

Figure 11 Figure 12 Figure 13
Figure 11. XRD spot intensity plotted as a function of vertical position in the 3 mm thick N5 bar casting. Spot intensity was high when the x-ray beam was directed into the solid single crystal, decreased in the mushy zone, and dropped to a background level when the x-ray beam passed through only molten alloy. Solidification model (Lever) predictions of the fraction of solid versus temperature are also plotted. Figure 12. Transmission topograph (top center) of dendrites in an N5 casting that are aligned with the [001] direction. The scale can be inferred by comparison with the diameter of the primary beam (large bright area; the actual size at the specimen is 3.5 mm). There is an additional vertical exaggeration caused by the topographic geometry. The dendrites appear to be approximately 1–2.5 mm wide. Figure 13. The geometry of transmission topography. X-rays reflected from lattice planes form an image of structure within the casting, which have a particular crystalline orientation.

Scans in the casting showed that the XRD spot intensity changed, as before, with vertical position in the casting. The topograph became less mottled and more solid when the x-ray beam was directed into the solidified crystal and had a more threadlike appearance higher in the casting where there was more molten alloy and less solid. The XRD spots, including the topograph, disappeared when the x-ray beam passed through only molten N5. The 001 topograph was best displayed during specimen rotations, where the three-dimensional character of the topograph could be appreciated.


Although we have used transmission XRD to sense the physical state (liquid, solid) of areas in a single-crystal casting, the orientation of the growing crystal is also easily measurable. Ordinarily, determining crystal orientation from one or two diffraction spots would be impossible; however, the information from several types of x-ray sensors can be combined to calculate the lattice plane spacing that produced the spot. The scattering angle 2 from the transmission XRD images and the energy of the diffracted x-rays with the energy-sensitive detector were measured. Thus, knowing the wavelength and scattering angle, the lattice plane spacing can be calculated. Since the specimen material and its crystal structure are known, indexing a particular XRD spot is straightforward. During several casting experiments, a change in the XRD pattern was observed as the x-ray sensor was scanned about the casting. This was a clear indication of a change in crystal orientation, which was confirmed after the experiment by etching the completed casting.


The authors are grateful to Steve Seltzer at NIST, Gaithersburg, Maryland, for providing a copy of the XCOM software. We also appreciate the financial support of this measurement technology development project from the NIST Office of Intelligent Processing of Materials and technical interactions with the other members of the NIST Aerospace Casting Consortium. In particular, we enjoyed the fruitful collaboration with the scientists and engineers at Howmet-Whitehall Casting (Boyd Mueller, George Strabel, John Brinegar, and Jeff King)–without whose help this project would not have been successful.

1. J.R. Helliwell et al., "Instrumentation for Laue Diffraction," Rev. Sci. Instr., 60 (7) (1989), pp. 1531–1536.
2. I.J. Clifton, M. Elder, and J. Hajdu, "Experimental Strategies in Laue Crystallography," J. Appl. Cryst., 24 (1991), pp. 267–277.
3. K. Harris, G.L. Erickson, and R.E. Schwer, "Directionally-Solidification and Single-Crystal Superalloys" Specialty Steels and Heat-Resistant Alloys (1982), pp. 995-1006.
4. R.E. Green, Jr., "An Electro-Optical X-Ray Diffraction System for Grain Boundary Migration Measurements at Temperature," Advances in X-Ray Analysis, vol. 15 (New York: Plenum Pub., 1972).
5. S.E. Doyle et al., "Probing the Structure of Solids in a Liquid Environment: A Recent In-Situ Crystallization Experiment Using High Energy Wavelength Scanning," J. Crys. Growth, 112 (1991), pp. 302–307.
6. N.R. Joshi and R.E. Green, Jr., "Continuous X-Ray Diffraction Measurement of Lattice Rotation during Tensile Deformation of Aluminum Crystals," J. Matls. Sci., 15 (1980), pp. 729–738.
7. O. Babushkin et al., "A High-Temperature Graphite Furnace for X-Ray Powder Diffraction," Meas. Sci. & Tech., 4 (1993), pp. 816–819.
8. K.G. Abdulvakhidov and M.F. Kupriyanov, "High-Temperature Unit for X-Ray Diffraction Studies of Single Crystals," Instr. & Exptl. Tech., 35 (5) (1992), pp. 942–943.
9. H.L. Bhat et al., "A Furnace for In-Situ Synchrotron Laue Diffraction and Its Application to Studies of Solid-State Phase Transformations," J. Appl. Cryst., 23 (1990), pp. 545–549.
10. S. Hosokawa, S. Yamada, and K. Tamura, "X-Ray Diffraction Measurements for Expanded Liquid Alkali Halides at Very High Temperature," J. Noncrys. Solids, 15 (1992), pp. 40–43.
11. R.E. Green, Jr., "Applications of Flash X-Ray Diffraction Systems to Materials Testing," Proc Flash Rad. Symp., ASNT Fall Conf. (1977) pp. 151–164.
12. C.J. Bechtoldt et al., "X-Ray Residual Stress Mapping in Industrial Materials by Energy Dispersive Diffractometry," Advances in X-Ray Analysis, vol. 25 (New York: Plenum Pub., 1981), pp. 329–338.
13. H.J. Kopinek, H.H. Otten, and H.J. Bunge, "On-Line Measuring of Technological Data of Cold and Hot Rolled Steel Strips by a Fixed Angle Texture Analyzer," Proc. 3rd International Symposium on Nondestructive Materials Characterization (New York: Plenum Pub., 1989), pp. 750-762.
14. R.D. Black et al., "Three Dimensional Strain Measurements with X-Ray Energy Dispersive Spectroscopy," J. Nondestr. Eval., 5 (1) (1985), pp. 21–25.
15. W. Riemers et al., "Evaluation of Residual Stresses in the Bulk of Materials by High Energy Synchrotron Diffraction," J. Nondestr. Eval., 17 (3) (1998).
16. J.H. Hubbell, H.M. Gerstenberg, and E.B. Saloman, Bibliography of Photon Total Cross Section (Attenuation Coefficient) Measurements 10 eV to 13.5 GeV, Internal Report NBSIR #86-3461 (Washington, D.C.: NBS, 1986).
17. B.E. Warren, X-Ray Diffraction (New York: 1990).
18. C.T. Chantler, "Theoretical Form Factor, Attenuation, and Scattering Tabulation for Z = 1–92 from E = 1–10 eV to E = 0.4–1.0 MeV," J. Phys. Chem. Ref. Data, 24 (1995), pp. 71–111.
19. D. Waasmaier and A. Kirfel, "New Analytical Scattering Factor Functions for Free Atoms and Ions," Acta Crystallogr. A, 151 (3) (1995), pp. 416-430.
20. M.J. Berger and J.H. Hubbell, XCOM: Photon Cross-Sections on a Personal Computer, Internal Report NBSIR #87-3567 (Washington, D.C.: NBS, 1987).
21. B.D. Cullity, Elements of X-Ray Diffraction, 2nd ed. (Publisher’s City: Addison Wesley, 1978), pp. 156–158
22. T.A. Campbell and J.N. Koster, "Visualization of Liquid-Solid Interface Morphologies in Gallium Subject to Natural Convection," J. Crys. Growth, 140 (1994), pp. 414–425.
23. L.H. Schwartz and J.B. Cohen, Diffraction from Materials, 2nd ed. (Berlin: Springer-Verlag, 1987).

D.W. Fitting, W.P. Dubé, and T.A. Siewert are employed at the Materials Science and Engineering Laboratory, Materials Reliability Division, National Institute of Standards and Technology.

For more information, contact T.A. Siewert, NIST, 325 Broadway, Boulder, Colorado, 80303; (303) 497-3523; e-mail siewert@nist.gov.

Copyright held by The Minerals, Metals & Materials Society, 1999

Direct questions about this or any other JOM page to jom@tms.org.

Search TMS Document Center Subscriptions Other Hypertext Articles JOM TMS OnLine