When compared to the typical configuration employed in transmitted light microscopy, the critical instrument parameters for reflected (or episcopic) light differential interference contrast (DIC) are much simpler, primarily because only a single birefringent Nomarski or Wollaston prism is required, and the objective serves as both the condenser and image-forming optical system. Because of the dual role played by the microscope objective, a Nomarski prism interference pattern projected into the objective rear focal plane is simultaneously positioned at the focal plane of the condenser illuminating lens system.
Reflected light microscopy is one of the most common techniques applied in the examination of opaque specimens that are usually highly reflective and, therefore, do not absorb or transmit a significant amount of the incident light. Slopes, valleys, and other discontinuities on the surface of the specimen create optical path differences, which are transformed by reflected light DIC microscopy into amplitude or intensity variations that reveal a topographical profile. Unlike the situation with transmitted light and semi-transparent phase specimens, the image created in reflected light DIC can often be interpreted as a true three-dimensional representation of the surface geometry, provided a clear distinction can be realized between raised and lowered regions in the specimen.
A schematic cutaway diagram of the key optical train components in a reflected light differential interference contrast microscope is presented in Figure 1. Illumination generated by the light source passes through the aperture and field diaphragms (not illustrated) in a vertical (episcopic) illuminator before encountering a linear polarizer positioned with the transmission axis oriented East-West with respect to the microscope frame. Linearly polarized light exiting the polarizer is reflected from the surface of a half-mirror placed at a 45-degree angle to the incident beam. The deflected light waves, which are now traveling along the microscope optical axis, enter a Nomarski prism housed above the objective in the microscope nosepiece where they are separated into polarized orthogonal components and sheared according to the geometry of the birefringent prism.
Acting in the capacity of a high numerical aperture, perfectly aligned, and optically corrected illumination condenser, the microscope objective focuses sheared orthogonal wavefronts produced by the Nomarski prism onto the surface of an opaque specimen. Reflected wavefronts, which experience varying optical path differences as a function of specimen surface topography, are gathered by the objective and focused on the interference plane of the Nomarski prism where they are recombined to eliminate shear. After exiting the Nomarski prism, the wavefronts pass through the half-mirror on a straight trajectory, and then encounter the analyzer (a second polarizer) positioned with the transmission axis oriented in a North-South direction. Components of the orthogonal wavefronts that are parallel to the analyzer transmission vector are able to pass through in a common azimuth, and subsequently undergo interference in the plane of the eyepiece fixed diaphragm to generate amplitude fluctuations and form the DIC image. Formation of the final image in differential interference contrast microscopy is the result of interference between two distinct wavefronts that reach the image plane slightly out of phase with each other, and is not a simple algebraic summation of intensities reflected toward the image plane, as is the case with other imaging modes.
A significant difference between differential interference contrast in transmitted and reflected light microscopy is that two Nomarski (or Wollaston) prisms are required for beam shearing and recombination in the former technique, whereas only a single prism is necessary in the reflected light configuration. Light passes through the same Nomarski prism twice, traveling in opposite directions, with reflected light DIC. The shear produced when the light waves pass through the prism on the way to the objective is cancelled during their second journey through the prism upon returning from the specimen surface. In this regard, the Nomarski prism and objective serve an identical function for incoming light waves as the first prism and condenser optical system in a transmitted light microscope. Similarly, light reflected from the specimen surface is gathered by the objective and focused into the Nomarski prism interference plane (conjugate to the objective rear focal plane), analogous to the manner in which these components function in transmitted light.
The optical pathway, both for the entire wavefront field and a single off-axis light ray, in reflected light DIC microscopy are illustrated in Figures 2(a) and 2(b), respectively. In each case, linearly polarized light from the polarizer is deflected by the half-mirror and enters the Nomarski prism located behind the objective. Sheared wavefronts are focused by the objective lens system and bathe the specimen with illumination that is reflected in the form of a distorted wavefront (Figure 2(a)) or the profile of an opaque gradient (Figure 2(b)) back into the objective front lens. The optical path difference produced between orthogonal wavefronts enables some of the recombined light to pass through the analyzer to form a DIC image. In Figure 2(b), note that the trajectory of the light ray incident on the specimen is displaced by the same distance from the microscope optical axis as the ray reflected from the surface.
The Wollaston and Nomarski prisms employed in reflected light DIC microscopy are fabricated in the same manner as those intended for use with transmitted light instruments. The single birefringent prism for reflected light is comprised of two precisely ground and polished wedge-shaped slabs of optical quartz that are identical in shape, but have differing orientations of the optical axes. In a Wollaston prism, the quartz wedges are cemented together at the hypotenuse with an orientation that positions the optical axes perpendicular to each other. Conversely, in a Nomarski prism, the axis of one wedge is parallel to the flat surface, while the axis of the other wedge is oriented obliquely. As a result of geometrical constraints, the interference plane for a Wollaston prism lies near the center of the junction between the quartz wedges (inside the compound prism), but the Nomarski prism interference plane is positioned at a remote location in space, outside the prism itself. Incident linearly-polarized light waves (parallel to the optical axis of the microscope) that enter a Wollaston or Nomarski prism are divided into two mutually perpendicular (orthogonal) components, termed the ordinary and extraordinary wave, which have identical amplitudes (70.7 percent of the original polarized wave) and are coherent (provided, of course, that the illumination source is also coherent). In order to produce orthogonal components having equal amplitudes, the linearly polarized light entering a Nomarski or Wollaston prism is oriented with the electric vector vibration direction positioned at a 45-degree angle with respect to the principal optical axis in the upper wedge of the prism.
An angular splitting or shear of the orthogonal wavefronts occurs at the boundary between cemented quartz wedges in a Wollaston prism, and the waves become spatially separated by an angle defined as the shear angle. At this boundary, the ordinary and extraordinary waves also exchange identities and diverge away from each other as a function of the refractive index experienced by each wave as it travels through the quartz prism. The shear angle and separation distance is constant for all incident wavefronts across the face of the prism, regardless of the entry point. The direction of wavefront shear is defined by the prism shear axis, which lies in the plane of the Wollaston prism and is parallel to the optical axis of the lower quartz wedge section. In a Nomarski prism, the wedge having an oblique optical axis produces wavefront shear at the quartz-air interface, and is responsible for defining the shear axis.
Nomarski and Wollaston prisms not only separate linearly polarized light into two orthogonal components, they also produce a relative phase shift (often termed an optical path difference) in each wavefront relative to the other. The degree of phase shift between the wavefronts varies linearly with the location of the input light beam in relation to the shear direction. Thus, the prism can be laterally translated along the optical axis of the microscope in the shear direction (a process known as introduction of bias retardation) to enable adjustment of the optical path difference introduced between the orthogonal wave components. In this manner, fine-tuning of the relative intensity in the image can be manipulated to produce the distinctive shadow-cast appearance for which DIC microscopy is so well known. Images appear as if they were illuminated from a highly oblique light source originating from a single azimuth.
Because the interference plane in a conventional Wollaston prism is positioned in the central portion of the prism, at approximately the centerline between the two quartz wedges, it is difficult to adapt this prism design for use with standard microscope objectives in reflected light DIC microscopy. This problem arises because the interference plane of the prism must coincide and overlap with the rear focal plane of the objective, which often lies below the thread mount inside a glass lens element. On the other hand, external displacement of the interference plane in Nomarski prisms renders them ideal for use with microscope objectives since they can be positioned some distance away (for example, in the nosepiece) and still establish a conjugate relationship between the objective rear focal plane and the compound prism interference plane. In a reflected light DIC microscope, the Nomarski prism is oriented so that the interference plane is perpendicular to the optical axis of the microscope (as is the objective rear focal plane).
The ordinary and extraordinary wavefronts proceeding to the specimen through a Nomarski prism experience optical path differences that have a magnitude dependent upon the location of the wave as it enters the prism. After the wavefronts exit the prism, they enter the objective lens system (acting as an illumination condenser) from the rear, and are focused into a parallel trajectory before being projected onto the specimen. Reflection of the orthogonal wavefronts from a horizontal, opaque specimen returns them to the objective, but on the opposite side of the front lens and at an equal distance from the optical axis (see Figure 2(b)). The waves gathered by the objective are focused on the Nomarski prism interference plane (again on the opposite side from their journey down), which results in a phase shift that exactly offsets the original difference produced before the waves entered the objective. As a result, the positional exchange of incident and reflected waves results in cancellation of relative phase shifts across the entire microscope aperture. A system of this type is referred to as being self-compensating, and the image produced has a uniform intensity. Compensation of the reflected light DIC system can be compared to that for transmitted light, where two matched, but inverted, Nomarski (or Wollaston) prisms are used to shear and recombine the beam. In the transmitted light configuration, the condenser prism (often termed the compensating prism) is imaged onto the objective prism (referred to as the principal prism) so that optical path differences are matched at every point along the surface of the prisms. Thus, in the transmitted light configuration, the principal and compensating prisms are separate, while the principal prism in reflected light DIC microscopy also serves the function of the compensating prism.
Light waves employed for reflected DIC microscopy must be at least moderately collimated in order to provide uniform compensation across the full beamwidth for the two required passes through the prism, and to insure that phase differences introduced by slopes and reflection boundaries in the specimen can be detected. Because the phase difference experienced by a beam on its first pass through the prism is governed by the pathway, accurate compensation of the reflected beam requires passage along a complimentary portion of the prism. In order to ensure collimation of the light beam, the microscope must be properly configured for Köhler illumination to guarantee that input waves are parallel (or nearly so) to the optical axis. A poorly collimated input beam will result in nonuniform compensation across the prism (and the resulting image), and destroys the unique phase relationship between orthogonal components at each image point.
Anatomy of the Reflected Light Microscope
The vertical illuminator is a key component in all forms of reflected light microscopy, including brightfield, darkfield, polarized light, fluorescence, and differential interference contrast. A reflected light (often termed coaxial, or on-axis) illuminator can be added to a majority of the universal research-level microscope stands offered by the manufacturers. The primary function of a vertical illuminator is to produce and direct semi-coherent and collimated light waves to the rear aperture of the microscope objective and, subsequently, onto the surface of a specimen. Reflected light waves gathered by the objective then travel a pathway similar to the one utilized in most transmitted light microscopes. A critical component of the vertical illuminator is a partially reflecting plane glass mirror (referred to as a half-mirror; see Figure 3) that deflects light traveling from the horizontal illuminator by 90 degrees into the vertical optical train of imaging components in the microscope. The half-mirror, which is oriented at a 45-degree angle with respect to both the illuminator and microscope optical axis, also allows light traveling upward from the objective to pass through undeviated to the eyepieces and camera system.
Modern vertical illuminators designed for multiple imaging applications usually include a condensing lens system to collimate and control light from the source. In addition, these illuminators contain an aperture iris diaphragm and a pre-focused, centerable field diaphragm to enable the microscope to operate in Köhler illumination (Figure 3). Vertical illuminators also have numerous slots and openings for insertion of light balancing and neutral density filters, polarizers, compensators, and fluorescence filter combinations housed in cube-shaped frames. When configured to operate with infinity-corrected objectives, vertical illuminators are equipped with a tube lens (see Figure 1) to focus light waves into the intermediate image plane. Both tungsten-halogen and arc-discharge lamphouses can be utilized with vertical illuminators (often interchangeably) to provide a wide range of illumination intensity and spectral characteristics.
The most popular choice of a light source for reflected light microscopy (including the DIC imaging mode) is the ubiquitous tungsten-halogen lamp, which features a relatively low cost and long lifespan. Incandescent halogen lamps are moderately bright, but require color balancing filters to raise their color temperature to daylight levels for digital imaging and traditional photomicrography with film. An alternative choice, useful at high magnifications and very low bias retardation values (where illumination intensity is critical), is the 75 or 150-watt xenon arc-discharge lamp. Xenon lamps feature a high level of brightness across the entire visible light spectrum, and have color a temperature output that approximates the value required for daylight balance.
When white light from a tungsten-halogen or arc-discharge lamp is used for illumination in reflected light DIC microscopy, the interference fringes associated with topographical changes in the specimen can actually appear in narrow rainbow patterns along the features as the various colors destructively interfere at slightly different locations on the surface. Use of a narrower wavelength band of illumination in specialized applications (for example, light emitted from a laser) will produce a DIC image where the fringes are established by the interference of a single wavelength. These fringes will be sharper and more defined, and their location will not depend upon the spectral response of the detector.
Illustrated in Figure 4 are images of the region near a bonding wire pad on the surface of a microprocessor integrated circuit captured in brightfield, darkfield, and differential interference contrast illumination using a vertical illuminator and reflected light. The brightfield image (Figure 4(a)) suffers from a significant lack of contrast in the circuit details, but provides a general outline of the overall features present on the surface. Darkfield illumination (Figure 4(b)) reveals only slightly more detail than brightfield, but does expose discontinuities near the vertical bus lines (central right-hand side of the image) and the bonding pad edges on the left. The differential interference contrast image (Figure 4(c)) yields a more complete analysis of the surface structure, including the particulate bonding pad texture, connections from the bonding pad to the bus lines, and numerous fine details in the circuitry on the left-hand side of the image.
Köhler illumination in reflected light microscopy relies on two variable diaphragms positioned within the vertical illuminator. The basic system is configured so that an image of the lamp filament is brought into focus at the plane of the aperture diaphragm, which is conjugate to the rear focal plane of the objective (where the filament can also be observed simultaneously in focus). A field diaphragm, employed to determine the width of the illumination beam, is positioned in the same conjugate plane as the specimen and the fixed diaphragm of the eyepiece. The aperture iris diaphragm is closer to the light source, while the field diaphragm is closer to the objective (the opposite configuration from that employed for transmitted illumination). Differential interference contrast is particularly dependent upon Köhler illumination to ensure that the waves traversing the Nomarski prism are collimated and evenly dispersed across the microscope aperture to produce a high level of contrast.
On most reflected light microscopes, the field diaphragm can be centered in the optical pathway by partially closing the iris aperture and translating the entire diaphragm via a set of centering screws (or knobs) adjacent to the aperture opening control lever. In practice, the field diaphragm should be opened until it is just outside the viewfield or the area to be captured on film or in a digital image. The primary purpose of the field diaphragm is to control the size of the field of view and to prevent stray light from obscuring specimen details. In conjunction with the field diaphragm, the aperture diaphragm determines the illumination cone geometry and, therefore, the angle of light striking the specimen from all azimuths. The iris diaphragm size can be modulated to adjust specimen contrast, and generally should be set to a size that is between 60 and 80 percent of the objective rear aperture. Such a setting provides the best compromise between maximum resolution and acceptable contrast.
Reflected Light DIC Microscope Configuration
Because the components for differential interference contrast must be precisely matched to the optical system, retrofitting an existing reflected light microscope, which was not originally designed for DIC, is an undesirable approach. Instead, all of the major microscope manufacturers now offer industrial and research-grade microscopes equipped with vertical illuminators and the necessary auxiliary optical components (usually marketed in kits) to outfit a microscope for DIC observation. Several different approaches to instrument design have yielded two alternatives for the introduction of bias retardation into the differential interference contrast microscope optical system.
The traditional method for establishing reflected light DIC is to employ a Nomarski prism attached to a mobile carriage within a rectangular frame (often termed a slider) that fits into the microscope nosepiece base, above the revolving objective turret (Figures 5(a) and 5(b)). In this design, bias retardation is introduced by rotating a thumbwheel positioned at the end of the slider that, in turn, translates the Nomarski prism back and forth laterally across the microscope optical axis. An alternative technique, termed de Sénarmont compensation (see Figure 6), utilizes individual fixed prisms for each objective (Figure 5(d)), and a quarter-wavelength retardation plate in combination with the linear polarizer (Figure 5(c)) to introduce an optical path difference (bias retardation) between orthogonal wavefronts.
All microscope designs that employ a vertical illuminator for reflected light observation suffer from the problem of stray light generated by the reflections from the illuminator at the surface of optical elements in the system. In particular, the upper and lower planar surfaces of the Nomarski prism can be problematic in producing annoying reflections that create excessive glare and degrade image quality. To counter this effect, Nomarski prisms designed for reflected light microscopy are fabricated so that the interference plane is positioned at an angle with respect to the shear axis of the prism (see Figure 2(b)). When the interference plane of the specialized Nomarski prism is brought into coincidence with the objective rear focal plane (perpendicular to the microscope optical axis) by its positioning inside the sliding frame or fixed housing, the flat outer wedge surfaces are now inclined with respect axial illumination pathway (Figures 1, 2(b), and 5(a)). As a result, reflections are diverted away from the half-mirror, specimen, eyepieces, and camera system so as not to adversely affect image intensity and contrast.
In modern microscopes, the distance between the objective focal plane and the seating face on the nosepiece is a constant value, often referred to as the parfocal distance. Therefore, a single Nomarski prism can often be mounted at a fixed distance from the objective seats (and rear focal planes) on the nosepiece in a slider frame, and service the entire magnification range with regards to beam shearing and recombination duties. In some cases, especially at the higher magnifications, variations in the position of the objective rear focal plane can be accommodated by axial translation of the Nomarski prism within the slider (illustrated in Figures 5(a) and 5(b)). This is often accomplished with a knob or lever that relocates the entire prism assembly up and down along the microscope optical axis. Reflected light microscopes that utilize a single prism for DIC are able to introduce bias retardation by laterally translating the prism across the microscope optical axis with a thumbwheel. The entire Nomarski prism slider can be removed from the optical path when the microscope is used for other imaging modes (brightfield, polarized light, darkfield, and fluorescence).
Microscopes equipped with a single translatable Nomarski prism in the nosepiece require only a polarizer and an analyzer as accompanying components in order to operate in differential interference contrast imaging mode. The polarizer is usually mounted together with a rack-and-pinion or planetary gearset into a thin rectangular frame, so that the transmission azimuth can be rotated through 360 degrees with a thumbwheel. The polarizer frame is introduced into the light path between the field diaphragm and the half-mirror through a slot in the vertical illuminator. Likewise, the analyzer can also be housed in a frame that enables rotation of the transmission axis. Analyzer frames are usually placed into a slot in the nosepiece or near the tube lens in the upper portion of the vertical illuminator. In some cases, either the analyzer or polarizer is mounted in a fixed frame that does not allow rotation, but most microscopes provide the operator with the ability to rotate the transmission azimuth of at least one of the polarizers in order to compensate for opaque specimens that absorb light. Housing the polarizer and analyzer in slider frames enables the operator to conveniently remove them from the light path for other imaging modes. When the polarizers remain in place and the Nomarski prism slider is removed, the microscope is configured for observation in polarized reflected light mode.
An alternative mechanism for introduction of bias retardation into the reflected light DIC microscope optical system is to couple a de Sénarmont compensator in the vertical illuminator with fixed-position Nomarski prisms (illustrated in Figures 5(c), 5(d), and 6) for the objectives. In the de Sénarmont configuration, each objective is equipped with an individual Nomarski prism designed specifically with a shear distance to match the numerical aperture of that objective. The prisms are glued into frames and housed in a dust-tight assembly that mounts between the objective and the microscope nosepiece (Figure 5(d)). Objectives are threaded into the Nomarski prism housing, which is then secured to the nosepiece. A small lever is used to shift the prism frame into and out of the optical pathway (the prism positionlever in Figure 5(d)). Because of the increased number of Nomarski prisms required for the de Sénarmont DIC microscope configuration, these accessories are considerably more expensive than the sliding prism in a traditional reflected light Nomarski DIC microscope.
The optical train of a reflected light DIC microscope equipped with de Sénarmont compensation is presented in Figure 6. Light from the illumination source is focused by the collector lens and passes through the aperture and field diaphragms before encountering a linear polarizer in the vertical illuminator. Positioned directly behind the polarizer in the optical pathway is a quarter-wavelength retardation plate fixed into position where the fast axis is oriented East-West with respect to the microscope frame. Together, the polarizer and retardation plate comprise the de Sénarmont compensator (Figure 5(c)). When the polarizer transmission azimuth is aligned parallel to the fast axis of the retardation plate in the de Sénarmont compensator, linearly polarized light emerges from the assembly, and is deflected at a 90-degree angle by the vertical illuminator half-mirror into the pathway of imaging elements in the microscope.
After the polarized light waves reach the half-mirror and are deflected, the remainder of the microscope optical train operates in a manner similar to that of a traditional DIC reflected light microscope. Thus, on the downward journey through the reflected light microscope, linearly polarized light first encounters the fixed Nomarski prism and is sheared according to the geometry of the prism wedges. After being focused by the objective lens elements and projected onto the opaque specimen, light is reflected back into the objective where it converges at the rear focal plane (coincident with the Nomarski prism interference plane). Sheared wavefronts are recombined at the prism interference plane and proceed to the analyzer, where components that are parallel to the transmission azimuth are passed on to the intermediate image plane. At the image plane, constructive and destructive interference occurs between wavefronts emerging from the analyzer to generate the DIC image.
Bias retardation is introduced into the reflected light de Sénarmont DIC system simply by rotating the linear polarizer in the vertical illuminator. By rotating the polarizer transmission azimuth with respect to the fast axis of the retardation plate, elliptically and circularly polarized light having an optical path difference between the orthogonal wavefronts is produced. When the polarizer axis is rotated up to 45 degrees in one direction, right-handed elliptical or circular polarizer light emerges from the de Sénarmont compensator. Rotating the polarizer in the opposite direction produces elliptical or circular wavefronts having a left-handed rotational sense. The optical path difference introduced by rotating the polarizer (over a range of plus or minus one-half wavelength) is further compounded when the orthogonal wavefronts enter the Nomarski prism and are sheared across the face of the prism. Introducing an optical path difference at the de Sénarmont compensator is analogous to the effect achieved when the objective Nomarski prism is translated across the optical path in a traditional DIC microscope configuration.
Characteristics of Reflected Light DIC Specimens and Images
In reflected light DIC microscopy, the optical path difference produced by an opaque specimen is dependent upon the topographical geometrical profile (surface relief) of the specimen and the phase retardation that results from reflection of sheared and deformed orthogonal wavefronts by the surface. For a majority of the specimens imaged with DIC, the surface relief varies only within a relatively narrow range of limits (usually measured in nanometers or micrometers), so these specimens can be considered to be essentially flat with shallow optical path gradients that vary in magnitude across the extended surface. Phase changes occurring at reflection boundaries present in the specimen also produce and optical path difference that leads to increased contrast in the DIC image. These phase differentials are more likely to be found at junctions between different media, such as grain boundaries and phase transitions in metals and alloys, or aluminum and metal oxide regions in a semiconductor integrated circuit.
Although reflected light DIC microscopy has been heavily employed for examination of metallographic specimens for the past few years, currently the most widespread and significant application is the examination of semiconductor products as a quality control measure during the fabrication process. In fact, most of the manufacturers now offer microscopes designed exclusively for examination of integrated circuit wafers in DIC, brightfield, and darkfield illumination. DIC imaging enables technicians to accurately examine large volumes of wafers for defects that are not revealed by other microscopy techniques (as illustrated in Figure 4). Minute variations in the geometrical profile of the wafer surface appear in shadowed relief, and maximum image contrast is achieved when the Nomarski prism setting is adjusted to render the background a neutral gray color.
When the Nomarski prism is translated along the microscope optical axis in a traditional reflected light DIC configuration, or the polarizer is rotated in a de Sénarmont instrument, an optical path difference is introduced to the sheared wavefronts, which is added to the path difference created when the orthogonal wavefronts reflect from the surface of the specimen. The net result is to render the specimen image in pseudo three-dimensional relief where regions of increasing optical path difference (surface relief or reflection boundaries) appear much brighter or darker, and those exhibiting decreasing path length appear in reverse. Distinguishing features on the specimen surface appear similar to elevated plateaus or sunken depressions, depending on the gradient orientation or reflection characteristics. Unlike the situation with transmitted light DIC, the three-dimensional appearance often can be utilized as an indicator of actual specimen geometry where real topographical features are also sites of changing phase gradients.
The shadow-cast orientation is present in almost every image produced by reflected light DIC microscopy after bias retardation has been introduced into the optical system. In addition, the direction of optical shear is obvious and can be defined as the axis connecting regions of the image displaying the highest and lowest intensity values. Surface features become distinguishable because shadow directions are often reversed for specimen details that posses either a higher or lower topographical profile than the surrounding surface. Because the shear axis is fixed by Nomarski prism design and other constrains involved in wavefront orientation for reflected light DIC microscopy, the axis direction cannot be altered to affect specimen contrast through a simple setting on the microscope. However, the relative phase retardation between sheared wavefronts can be reversed by relocating the Nomarski prism from one side of the microscope optical axis to the other (shifting the bias retardation value from negative to positive, or vice versa). The same maneuver can be accomplished by rotating the polarizer to the corresponding negative value on a de Sénarmont compensator. When phase retardation is altered as just described, the orientation of bright and dark edges in the image is reversed by 180 degrees.
Orientation Effects on Reflected Light DIC Images
As discussed above, reflected light DIC images are inherently bestowed with a pronounced azimuthal effect, which is the result of asymmetrical orientation of the beamsplitting Nomarski prism with respect to the microscope optical axis and the polarizers. The result is that many opaque specimens imaged in differential interference contrast have a prerequisite orientation limitation in order to achieve maximum contrast (either parallel or perpendicular to the shear axis) that restricts freedom of specimen rotation. This is especially critical with highly ordered semiconductors having numerous extended, linear regions intermixed with closely-spaced periodic structures.
Azimuth contrast effects in reflected light differential interference contrast can be utilized to advantage by equipping the microscope with a 360-degree rotating circular stage. An essential element in polarized light microscopy, circular stages enable the operator to rotate the specimen with respect to the shear axis in order to maximize or minimize contrast effects for selected specimen features. Contrast in reflected light DIC microscopy achieves a minimum level for linear phase specimens that extend along the direction of shear, but can be varied significantly by rotating the stage by 90 degrees. Non-linear metallurgical specimens, such as mosaic grain boundaries, wires, amorphous alloys, and crystalline spherulites, do not display significant azimuthal effects in reflected light DIC, and can usually be imaged satisfactorily in a variety of orientations.
Presented in Figure 7 are two semiconductor integrated circuit specimens, each having a significant amount of periodicity, but displaying a high degree of asymmetry when imaged in reflected light DIC. Figures 7(a) and 7(b) illustrate the same region of a microprocessor arithmetic logic unit located near the pad ring, which contains numerous bus lines, bonding wire pads and registers. When the circuit is positioned with the long axis of the bonding pad oriented perpendicular to the shear axis (northwest to southeast for all images in Figure 7), the central trapezoid-shaped region of bus lines becomes very dark and almost extinct (Figure 7(a)), losing virtually all recognizable detail. Rotating the integrated circuit by 90 degrees (Figure 7(b)), highlights the central trapezoid bus structure, but causes adjacent areas to lose contrast. In order to capture all the detail present on the surface of this integrated circuit, the optimum orientation is to position the elongated bus structure at a 45-degree angle to the shear axis of the microscope.
Several mask alignment markers are illustrated in the image of a semiconductor surface presented in Figure 7(c). One of the markers has been placed on a metallic bonding pad, while the other rests on a smooth metal oxide surface. Both markers contain eight lines, equally spaced at 45-degree intervals, and having the same length. Note that the lines oriented parallel to the shear axis are very dark for the marker resting on the metal oxide surface, while the upper left line on the bonding pad marker is almost invisible. The marker lines oriented perpendicular (northeast to southwest) to the shear axis are much brighter and far more visible than lines having other orientations, although the lines parallel and perpendicular to the image boundaries are clearly visible.
The correlation between image contrast and specimen orientation in reflected light DIC microscopy can often be utilized to advantage in the investigation of extended linear structures (especially in semiconductor inspection). By capturing images at several orientations, DIC microscopy is often able to present a clear representation of the complex morphology present in many extended, linear specimens. In addition, when optical sectioning methodology is coupled to azimuth-specific imaging, reflected light DIC microscopy can often reveal features that are difficult, or impossible, to distinguish using alternative techniques.
Optical Staining in Reflected Light DIC
Bias retardation between the sheared wavefronts in reflected light DIC microscopy can be manipulated through the use of compensating plates, such as a first-order (often termed a full-wave or first-order red) plate having a retardation value equal to a full wavelength in the green region (550 nanometers) of the visible light spectrum. Compensating plates bestow greater control for adjusting the contrast of specimen details in relation to the background intensity and color values, and also enable more precise tuning of the bias value between orthogonal wavefronts. These birefringent components are also frequently employed for optical staining of opaque specimens, which are normally rendered over a limited range of grayscale values.
Optical staining is accomplished either through translation of the Nomarski prism across the optical pathway by a significant distance from maximum extinction, or by inserting a full-wave compensator behind the quarter-wavelength retardation plate in a de Sénarmont configuration. A full range of interference colors can be observed in specimen details when the Nomarski prism is translated to extreme ranges, or the polarizer is rotated with de Sénarmont compensation coupled to a full-wave plate. With the compensator in place, the background appears magenta in color, while image contrast is displayed in the first-order yellow and second-order blue colors of the Newtonian interference color spectrum. Under these conditions, small variations in bias retardation obtained by translation of the Nomarski prism (or rotating the polarizer in a de Sénarmont compensator) yield rapid changes to interference colors observed in structures having both large and small surface relief and reflection phase gradients.
Illustrated in Figure 8 are three specimens imaged in reflected light DIC with a full-wave retardation plate inserted behind the de Sénarmont compensator in a fixed-prism microscope configuration. Figure 8(a) reveals surface defects in a high-temperature superconducting ceramic single crystal grown from an amorphous flux at 950 degrees Celsius. Although twinning defects in the crystal are difficult to discern without applying optical staining techniques, these crystalline mishaps become quite evident and are manifested by significant interference color fluctuations when the retardation plate is installed. Similarly, adhesion failure in a magnetic thin film is clearly imaged when optical staining techniques are employed in reflected light DIC (Figure 8(b)). Separation points in the film are imaged as wrinkles that appear in spectacular relief, surrounded by interference fringes, when observed in white light. Finally, bus line details stand out in sharp color contrast on the surface of the integrated circuit presented in Figure 8(c). A wide spectrum of differential color effects are possible with integrated circuits in reflected light DIC microscopy, based on a number of factors, including the presence or absence of silicon nitride or polyimide protective coatings, phase relationships between fabrication materials, and the feature linewidth of the fabrication process.
Although optical staining is also possible in transmitted light DIC, the effect is far more useful with reflected light techniques, especially when examining flat, planar specimens, such as integrated circuits that have surface relief variations restricted to relatively narrow limits. In contrast to the transparent specimens imaged with transmitted light, surface relief in opaque specimens is equivalent to geometrical thickness. In addition, localized differences in phase retardation upon reflection of incident light from an opaque surface can be compared to the refractive index variations experienced with transmitted light specimens. For many applications in reflected light DIC, specimen details are frequently superimposed on a homogeneous phase background, a factor that dramatically benefits from contrast enhancement through optical staining (interference) techniques. With the thin transparent specimens that are optimal for imaging with transmitted light DIC, the range within which optical staining can be effectively utilized is considerably smaller (limited to a few fractions of a wavelength), rendering this technique useful only for thicker specimens.
Optical Sectioning in Reflected Light DIC
The ability to capitalize on large objective numerical aperture values in reflected light DIC microscopy enables the creation of optical sections from a focused image that are remarkably shallow. Without the confusing and distracting intensity fluctuations from bright regions occurring in optical planes removed from the focal point, the technique yields sharp images that are neatly sliced from a complex three-dimensional opaque specimen having significant surface relief. This property is often employed to obtain crisp optical sections of individual features on the surface of integrated circuits with minimal interference from obscuring structures above and below the focal plane.
In reflected light microscopy, the vertical illuminator aperture diaphragm plays a major role in defining image contrast and resolution. Reducing the aperture size increases the apparent depth of field and overall image sharpness while simultaneously producing enhanced contrast. However, if the diaphragm is closed too far, diffraction artifacts become apparent, image intensity is significantly reduced, and resolution is sacrificed. Often, the optimum aperture diaphragm setting is a compromise between accurately rendering specimen detail in sufficient contrast and retaining the resolution necessary to image minute features, while at the same time avoiding diffraction artifacts.
The series of high-magnification DIC images presented in Figure 9 illustrate three separate focal planes in the same viewfield of overlapping surface structures present on a typical integrated circuit. In brightfield or darkfield illumination, these structures are often observed merged together and can become quite confusing when attempting to image specific surface details. Figure 9(a) reveals several metal oxide terminals on the upper surface of the integrated circuit, including vias (miniature connections between vertical layers) and part of a bus line. Refocusing the microscope a few tenths of a micrometer deeper exposes numerous connections in the central region of the circuit (Figure 9(b)). Still farther into the circuitry, near the first layers applied above the pure silicon, are a series of metal oxide lines dotted with an ordered array of via connections (Figure 9(c)). The optical sectioning capability of reflected light DIC microscopy is clearly revealed by the ability to image specific focal planes on the surface of this complex integrated circuit.
An essential feature of both reflected and transmitted light differential interference contrast microscopy is that both of the sheared orthogonal wavefront components either pass through or reflect from the specimen, separated by only fractions of a micrometer (the shear distance), which is much less than the resolution of the objective. To the observer, it is not apparent that the resulting image visualized in the eyepieces is composed of these two superimposed components, because their separation is too minute to be resolved by the microscope. However, each point in the image is derived from two closely spaced and overlapping Airy disks originating from adjacent points on the specimen, and each disk has an intensity that corresponds to its respective optical path difference induced by the specimen. Image contrast is described as being differential because it is a function of the optical path gradient across the specimen surface, with steeper gradients producing greater contrast.
Primary candidates for observation in reflected light DIC microscopy include a wide variety of metallographic specimens, minerals, alloys, metals, semiconductors, glasses, polymers, and composites. The high resolution afforded by the technique has been employed to ascertain specimen details only a few nanometers in size. For example, spiral growth dislocation patterns in silicon carbide crystals that are only about 30-40 nanometers high can be imaged in high relief, while thin films approximately 200 nanometers thick have been successfully observed in monochromatic yellow sodium light. Phase transitions and recrystallization processes can be examined in reflected light DIC, as well as minute details on the surface of glasses and polymers.
Although largely a tool restricted to industrial applications, reflected light differential interference contrast microscopy is a powerful technique that has now been firmly established in the semiconductor manufacturing arena. The millions of computer chip components fabricated each year rely heavily on reflected light DIC to ensure quality control and help prevent failure of the circuits once they have been installed. Because of the countless hours spent by technicians examining integrated circuits, microscope manufacturers are now carefully turning their attention to ergonomic considerations in the design of new reflected light instruments. The result will undoubtedly be highly refined microscopes that produce excellent DIC images, while minimizing the discomfort and neuro-muscular disorders experienced by operators who must spend long periods repetitively examining identical specimens.
Chris Brandmaier - Industrial Microscope Division, Nikon Instruments, Inc., 1300 Walt Whitman Road, Melville, New York 11747.
Kenneth R. Spring - Scientific Consultant, Lusby, Maryland, 20657.
Michael W. Davidson - National High Magnetic Field Laboratory, 1800 East Paul Dirac Dr., The Florida State University, Tallahassee, Florida, 32310.