Scanning Microscopy Moves into Production
Scott Jordan, PI USA
Since the scanning tunneling microscope was invented in the early 1980s, the field of microscopic imaging has blossomed in many nontraditional ways. Nearly 400 years of steady evolution of optical microscopy and five decades of progress in the field of electron microscopy failed to prepare the technical world for the avalanche of new techniques.
Today scanning probe microscopy encompasses a wide variety of profiling methodologies that leverage physical interactions to characterize surface topographies in novel ways. In the words of James K. Gimzewski, a pioneer in the field, the principle is now the foundation for a “laboratory on a tip.” In essence, all manner of surface physics properties are exploited to create ever more detailed images of the unseen world of surfaces and molecules and to illuminate the behavior and characteristics of materials in revolutionary ways.
Equally important, scanning probe microscopy is an established methodology for the inspection and monitoring of production processes. For example, in the semiconductor industry, thin-film quality measurement and metrology of a microlithographic structure’s critical dimensions are routine tasks in the production of integrated circuits.
Prominent among these applications is classical atomic force microscopy, which observes the Van der Waals force arising from the interaction of the surface with a very fine probe as one is scanned vs. the other. Performing a raster scan of the sample compiles a detailed image of the surface’s atomic-scale topography. Other tools apply force with the stylus during the scanning process to characterize hardness, frictional coefficients and other critical material properties.
Although scanning probe microscopy has always required extraordinary positioning resolution, the emergence of industrial applications for the technique has placed extreme demands on trajectory control and system speed as well. Piezo tube actuators are sometimes employed to scan the probe, but these devices have limited travel and a curved trajectory, which makes them inappropriate for some uses.
To address these issues, flexure-guided X-Y and X-Y-Z stages have been developed that combine long travel with the traditionally high resolution of piezo actuation. Sensitive position sensors are usually integrated into these stage mechanisms. Design selections include strain gauges such as piezoresistive sensors, linear variable differential transformers and capacitive sensors.
Until recently, capacitive sensors tended to be the costliest choice. Recent advances have helped lower the cost of these sensitive, stable and high-bandwidth-feedback devices. They have the significant advantage of continuously measuring the position of the moving platform — in phase with the motion — to a precision of angstroms, rather than inferring position from flexure or piezo stack strain. This improves accuracy, repeatability and phase fidelity, and it results in crisp, high-bandwidth motion — critical considerations in industrial applications, where time is money.
In addition, the noncontact metrology facilitated by capacitive sensors has freed designers to devise parallel-kinematics/-metrology mechanisms whereby a single platform is actuated and monitored simultaneously in multiple degrees of freedom. In this way, transverse runout in the X-axis, for example, is automatically and continuously compensated by the Y-axis mechanism. This feat is impossible using serial metrology feedback and stacks of independent motion axes, which are also slower in actuation because the bottom axis bears all the mass of the upper axes.
Because the scanning stage serves as a datum plane for the metrology, any out-of-plane defect in the stage’s motion will appear as an artifact in the scan that may be indistinguishable from details on the sample’s surface. This has driven the development of novel flexure mechanisms in which every detail of flexure configuration and manufacture is optimized for planarity of motion. As a result, it is now possible to purchase off-the-shelf parallel-kinematics nanopositioners with planarity to 10 nm over scanning ranges of 100 × 100 μm or even better with active Z actuation.
Controller advancements have also played a role in advancing scanning probe microscopy into production applications. New techniques such as PI’s dynamic digital linearization virtually eliminate following error and nonlinearity in rapid, repetitive scanning. Another technique, called Input Shaping, eliminates recoil-generated ringing throughout the load and structure, improving resolution and reducing spurious artifacts in the image (see figure).
During the scan of a J-shaped semiconductor resolution target, approximately 2 μm wide, vibration related to high acceleration in the piezo stage can produce banding distortion in the image (top). Using PI’s Input Shaping routine in the controller, it is possible to eliminate the artifacts caused by the vibration (bottom).
In the future, scanning probe microscopy will play an increasingly important role in fields as diverse as production process control and drug screening. Advancements in nanopositioning mechanics and controls will continue to reflect the innovation driving these applications.
Meet the author
Scott Jordan is director of nanopositioning technologies at PI USA (Physik Instrumente) LP in Auburn, Mass.; e-mail: firstname.lastname@example.org.
MORE FROM PHOTONICS MEDIA