Atomic-force microscopy guides industrial manufacturing and R&D

F. MICHAEL SERRY

Very often, the performance obtained from an optical element is only possible with the thin film(s) that coat the element. From camera lenses to microlithography masks, from contact lenses to micromirrors in optical MEMS, thin-film coatings play important, often critical roles in how these elements interact with light and with their environment to elicit well-defined, as-designed performance.

Three-dimensional atomic-force microscope (AFM) "height" images provide high-resolution topographic maps of the surface that help us understand the detailed structure of optical thin films and substrates. With that understanding, we can make better decisions about the composition or deposition conditions of a thin film, or the polishing procedure to smooth out a substrate on which a thin film is deposited, all with the aim of obtaining a desired optical performance from a component.

Each AFM image contains a wealth of information that can be quantified in numerous ways. For example, power-spectral-density (PSD) analysis in AFM software quantifies surface texture beyond root-mean-square (rms) roughness and other wavelength-indiscriminant numbers. A PSD analysis on a typical AFM image shows the contribution to the surface roughness at different in-plane wavelengths, from tens of micrometers down to a few nanometers, even angstroms—far below the wavelength of visible light. This information can help determine where in the manufacturing process to search for ways of improving the performance of a component for a given application.

To reduce scatter losses from a fluoride-based thin-film coating on a quartz substrate, for example, the results of the PSD analysis on AFM images captured from the surfaces of the bare and the coated substrates can be compared directly to determine which surface—the substrate or the coating—contributes more to the scatter losses at or near a given wavelength. In one case study conducted at the Fraunhofer Institute of Applied Optics and Precision Engineering (Jena, Germany), PSD analysis of AFM images showed that a single-layer thin-film coating did not alter the scattering properties from those of the uncoated substrate at visible and ultraviolet wavelengths (see Fig. 1).¹ This meant that any improvement in the scattering performance of the substrate covered with this single-layer thin-film coating most likely required improved polishing of the substrate. On the other hand, the multilayer thin-film coating on the same type of substrate was clearly the dominant contributor to scattering losses across all in-plane wavelengths measured, and the further polishing of the substrate was deemed not nearly as important in this application as in the single-layer thin-film application.

Contact lens in saline solution

An AFM image showing the morphology of a thin-film surface submerged in a liquid is a uniquely useful resource for thin-film engineering in many areas of science and technology, notably in biomaterials. The AFM is the only instrument that combines nanometer- and angstrom-scale resolution in three dimensions with in-liquid operation on a very wide range of materials, including polymers. In-liquid operation is critical for studying thin films whose native environment in the final product is either inside the human body (such as implants) or in immediate contact with mucous membranes.

In contact lenses, for instance, deposition of a thin-film coating, along with subsequent industrial processing, is a method intended to make an originally hydrophobic polymer surface hydrophilic for improved biocompatibility in the eye of a patient. AFM images can assist this process by showing the surface morphology in saline solution of an experimental contact lens before and after processing. In one instance, the depressions in the image of the hydrophobic polymer substrate were a surprise finding and were considered defects, which the manufacturer was able to eliminate after experimenting with and altering the manufacturing process (see Fig. 2).

FIGURE 2. AFM images of contact lenses in saline solution before and after thin-film coating and processing (left and right respectively) enabled the contact-lens manufacturer to better control the manufacturing process for improved biocompatibility. Both images are 8-µm on a side.

Visualizing and measuring the in-plane distribution and size of the protrusions (lighter in color scale) in the thin-film-coated surface (Fig. 2, right) and comparing with the image of the uncoated lens hinted that the depressions on the uncoated substrate may have caused the thin-film coating to aggregate locally, resulting in the highly textured surface post-processing. The surface of this experimental lens material looks remarkably different when imaged with the AFM in air—both the bare and the thin-film-coated material—so much so that the findings taken from the in-liquid images were impossible to make from the images captured in air.

Beyond purely optical elements

The electrical properties of thin-film transistors (TFT) in liquid-crystal displays (LCD) can vary with the properties—such as surface texture—of the thin-film silicon, which is the electrically active material in the TFT. For example, the average grain size and distribution of polycrystalline silicon (p-Si) affects the mobility of charge carriers in the TFT in active-matrix liquid-crystal-display (AMLCD) panels. This variation must be well characterized, measured, and controlled in order that the performance of each pixel in the LCD is satisfactory, and similar to that of the other pixels in the display. Here, morphology uniformity across different size areas takes on elevated importance, as well as the details of that morphology within a given size area. The AFM is a powerful analysis tool for studying grain size and distribution.

One method to make p-Si from amorphous Silicon (a-Si) is to use an excimer laser beam to melt the a-Si, which then is allowed to cool down to form the granular p-Si (see Fig. 3). Variation of p-Si grain morphology depends on how a-Si crystallizes after laser melting (which may also depend on the morphology of the starting a-Si). Transformation of a-Si into p-Si depends in part on the laser beam and laser pulse characteristics, including beam shape and pulse duration. Conversely, detailed, quantifiable information about the silicon grain structure can help characterize the laser beam and pulse, and fine-tune it so as to achieve the desired grain surface morphology.

FIGURE 3. AFM images show the surface morphology of same-size areas on a silicon thin film deposited on a glass substrate before and after excimer laser annealing—a-Si (left) and p-Si (right), respectively. The images are 2 µm on a side and the height (z) scale is the same for both, with lighter shades of blue standing for taller features. The areas colored in red are all at or above a given user-selectable height, and define grain size and distribution at that selected height level.

A composite of four separate AFM images that cover a contiguous area shows gradually changing grain size, height, and distribution along one direction—the same direction along which the laser beam was scanned across the sample to transform a-Si into p-Si (see Fig. 4). Such variations can be quantified in numerous analysis modules that are standard features on most commercially available AFMs.

FIGURE 4. Grain size and distribution varies across this 20 x 50-µm area of laser-annealed silicon. These variations can be quantified with AFM analysis software to help guide process engineers in fine-tuning the process parameters to obtain optimal surface texture.

We can select a height level from which to start grain-size analysis, for example. This results in a map of grains created by virtually cross-sectioning the image at that height level (red regions in Fig. 3). The grain distribution is then quantified, showing the average grain size, numerating grains of any given size (within a given tolerance) and so on. We can then decide to apply filters to the grain-size image to either separate groups of grains that are marginally conjoined so as to count them as multiple, smaller grains (this is called grain erosion); or to join into fewer, larger grains groups of several grains that are marginally separated (this is called grain dilation).

These options help define and measure acceptable grain sizes, and thus define process windows that better control grain size and uniformity. This is all possible because an AFM image provides a three-dimensionally quantifiable map of the surface of the sample it images.

REFERENCE

Veeco Intruments Application Note, Using Atomic Force Microscopy for Engineering Low Scatter Thin Film Optics, A. Duparre, N. Kaiser (Fraunhofer Institute of Applied Optics and Precision Engineering; Jena, Germany) and M. G. Heaton (Veeco Instruments).

F. MICHAEL SERRY is senior applications scientist at Veeco Instruments, 112 Robin Hill Road, Santa Barbara, CA 93117; e-mail: [email protected].