Waters Symposium: X-ray Diffraction of Powders and Thin Films
The Development of Instrumentation for Thin-Film X-ray Diffraction Tom Ryan† Philips Analytical, 12 Michigan Drive, Natick, MA 01760
Introduction to Thin-Film X-ray Diffraction The penetration depth of Cu Kα X-rays in typical solid materials is a few tens of microns. This characteristic alone makes X-ray diffraction a natural choice for structural characterization of thin surface films. Today, XRD is a standard method in laboratories and production environments to measure the structural properties of thin films. Thin-film XRD has been instrumental in enabling the development of a diverse range of applications from lasers for fiber-optic telecommunications through high-density magnetic recording heads to surface modification coatings for artificial hip joints. In many cases, thin-film XRD is used in manufacturing plants as an on-line process control tool to control the composition and thickness of films only a few nanometers thick. It all seems very routine today. However, as recently as the late 1980s, thin-film XRD in the laboratory was very far from routine. Remarkably, the progress that has been made in applying XRD to ultrathin films has been made with few, if any, fundamental technological breakthroughs in X-ray diffractometry. Rather, we have seen a development of “understanding what is possible” and an expanding demand for thin-film characterization driven by the increasing importance of thin-film technology in many high-tech industries. Interestingly, this has led to a new generation of X-ray diffractometers, from all of the major manufacturers, which blur the distinction between the traditional powder diffractometer and the highly specialized thin-film diffractometers that were common a decade ago. Ron Jenkins has given an excellent review of powder diffraction and explained the fundamentals of powder diffractometry (1). In general, a “powder” sample comprises an infinitely thick tablet of polycrystalline material consisting of randomly oriented crystallites ranging in size up to a few tens of microns. Thin films, on the other hand, come in three distinct “flavors”: polycrystalline, amorphous, and epitaxial (or single crystal). In reality, the three tend to get mixed together to form an infinitely variable cocktail of properties. That’s one of the delights—and one of the problems—of this science. A polycrystalline film is, in essence, a powder sample of limited thickness deposited on a substrate. The substrate can be polycrystalline (e.g. surface coatings on tool steel), it can be amorphous (e.g. antireflective coatings on window glass), or it can be a single crystal (e.g. metal interconnects in Si ICs). In each case, the challenge is to maximize the XRD signal from the thin polycrystalline film while suppressing the XRD signal from the substrate. Amorphous thin films present a different set of problems. Once again, they may be found on any of the three substrates. Perhaps the most common example of an amorphous film is SiO2, which is used as a dielectric (insulator layer) in silicon ICs. Amorphous films have no long-range crystalline structure. † Present address: Emcore Corporation, 145 Belmont Dr., Somerset, NJ 08873.
They don’t diffract X-rays. But an X-ray diffractometer can be used, very successfully, to measure the thickness of an amorphous film by a technique known as X-ray reflectivity. Epitaxial thin films, by contrast, diffract X-rays very strongly indeed. An epitaxial thin film is a single-crystal film grown on a single-crystal substrate. The substrate acts as a template for the film growth, so, for obvious reasons, epitaxial films are not found on polycrystalline or amorphous substrates. Epitaxial films are found in solid-state lasers (for CD players or fiber optics), in light-emitting diodes (LEDs), and in high-speed electronic devices. For example, digital cell phones rely on an epitaxially grown chip to power the antenna. When Is a Thin Film Thin? This is a very pertinent question that lies right at the heart of the story. As I mentioned at the start of my discussion, the penetration depth of an X-ray beam is typically a few to a few tens of microns. In fact, polycrystalline films in this thickness range are quite easy to deal with using a regular powder diffractometer. Below one micron, the signal from the film can be seriously obscured by the signal from the substrate. At 100 nm, the signal from the film may be only 1% of the signal from the substrate, and that approaches the limit of sensitivity of powder diffraction. Something must be done to enhance the signal from the layer and suppress the substrate signal. As we reach 10 nm things only get worse. Not only is the signal a factor of ten smaller, but diffraction broadening makes the diffraction peaks wider. The peak intensities drop rapidly. In laboratory XRD terms: 10 µm is a film 1 µm is a thin film 100 nm is a very thin film 10 nm is an ultrathin film 1 nm is a surface
Of course, it is perfectly possible to study atomic monolayers by XRD. But those are difficult experiments, normally carried out using synchrotron X-ray sources, and are beyond the scope of this discussion. In the 1980s the common belief was that laboratory XRD ran out of steam for sub-100-nm polycrystalline films. Imagine our consternation when, in 1987, one of our competitors (Rigaku) published an advertisement with the headline “Throwing Light on 100 Å” and showing an XRD measurement from a 10-nm metal film! In fact, the film was an “easiest case” example, a gold film on a silicon wafer. Nonetheless, this was really the beginning of serious thin-film diffraction on a commercial laboratory diffractometer. At the same time, Philips Research Laboratories were intensively involved in epitaxial film growth of compound semiconductors (InP and GaAs) for solid-state laser applications. These efforts relied heavily on XRD for characterization of the very thin epitaxial layers common at that time. The laboratories used a novel X-ray diffractometer developed by one of
JChemEd.chem.wisc.edu • Vol. 78 No. 5 May 2001 • Journal of Chemical Education
613
Waters Symposium: X-ray Diffraction of Powders and Thin Films
Polycrystalline Thin-Film XRD If you recall the X-ray optics of a conventional Bragg– Brentano X-ray diffractometer (3), you will note that the Xray beam is incident at an angle of half the diffraction angle (2θ) (Fig. 1). Typically, this is between 15° and 50°. The depth of penetration into the sample is given by the absorption length times the sine of the incidence angle. Since most of the diffraction peaks from a typical sample material lie in the region of 20o to 100° 2θ, incidence angles are typically in the range of (30 ± 20)°. If the sample is a thin layer, most of the X-ray beam passes through the film and is scattered by the substrate. The trick in thin-film XRD is to choose an X-ray diffraction geometry that lets you work at very small angles of incidence, increasing the path length of the X-rays in the film and reducing the amount of X-rays that penetrate through to the substrate. An incidence angle of 6° reduces the penetration depth by a factor of 10; an incidence angle of 0.6° leads to reduction by a factor of 100. X-ray diffraction at very shallow angles of incidence is called glancing incidence XRD. The optical path is shown in Figure 2. To define the incidence angle very accurately, the incident beam is collimated by a narrow divergence slit. During a measurement the incidence angle is fixed and the detector is scanned over the range of diffraction angles. In contrast to the Bragg–Brentano case, in which the diffracted beam is focused back onto a narrow receiving slit, the beam that emerges from the thin-film geometry is broad and close to parallel. The thin-film diffractometer uses a parallel-plate collimator to measure the angle of the diffracted X-ray beam (the Bragg angle). The collimator is made of a set of closely spaced metal plates, about 5 inches long, with an acceptance angle of around 0.25°. In fact, this collimator is a more or less standard component of an X-ray fluorescence spectrometer. A number of instrumental variations on the above theme are possible. For example, collimation and monochromation of the incident X-ray beam by reflection from a parabolic X-ray mirror reduces the background, increasing the sensitivity of the experiment. The addition of a diffracted-beam monochromator can also be used to reduce the instrumental background. Figure 3 illustrates the typical gain in sensitivity achieved using a thin-film diffractometer. Epitaxial Thin-Film XRD X-ray diffraction from an epitaxial layer structure could not be more different from the polycrystalline case. In the 614
Figure 1. Bragg–Brentano diffraction geometry.
Figure 2. Thin-film diffraction geometry.
500
Glancing incidence XRD: 2 degree incidence angle
400
Counts
my colleagues, Wim Bartels (2). It was obvious that thin films were “the next big thing”. The opportunity was recognized and a team was formed to develop a commercial thin-film X-ray diffractometer. That initiative led to the creation of the Philips MRD (Materials Research Diffractometer), which today is one of the mainstays of Philips XRD business. At the same time, similar processes were taking place within other XRD companies. In the Siemens Research laboratories the group of Herbert Goebel was very active in developing thinfilm XRD capabilities and, in particular, contributed to the rediscovery of X-ray reflectivity for thickness measurement of ultrathin films. In England, the groups of Keith Bowen and Brian Tanner at Bede Scientific focused on epitaxial thin films. I’m sure that the Philips story is not too different from the Siemens, Bede, and Rigaku stories!
300 200 100 0 5000
Conventional Bragg-Brentano XRD
0 20
30
40
50
60
70
2θ / deg Figure 3. Thin-film and conventional XRD. The sample is a thin film of metal on glass.
epitaxial structure, both the layer and the substrate are almost perfect single crystals. That means that they diffract only at very well defined angles of incidence and reflection and then only over a very tiny range of angles (a few seconds of arc). In general, the crystal structure of the layer is almost identical to that of the substrate, so that the difference in diffraction angle between layer and substrate is also very small. If the layer is thin (less than 100 nm), the layer diffraction peak will be broadened. As it gets thinner, it broadens and weakens. The challenge, then, is to separate two closely spaced diffraction peaks, one (the substrate) being very intense and very narrow and the other (the layer) being very weak and broad. Epitaxial thin film diffraction is sometimes called high-resolution X-ray diffraction. This term is related to the need to resolve Bragg peaks on an arc second scale. For an excellent review see the textbook by Bowen and Tanner (4 ). Figure 4 shows a typical XRD measurement from an epitaxial thin film. In fact this is a thin layer of SiGe (an alloy of
Journal of Chemical Education • Vol. 78 No. 5 May 2001 • JChemEd.chem.wisc.edu
Waters Symposium: X-ray Diffraction of Powders and Thin Films
Figure 4. X-ray rocking curve of epitaxial thin film.
Figure 5. Schematic of a Bartels high-resolution diffractometer.
from perfection decreases the resolving power of the diffractometer and increases the background from the very strong substrate Bragg peak, potentially obscuring the very weak X-ray diffraction signal from the thin epitaxial layer. The breakthrough came in the form of an X-ray monochromator developed by Wim Bartels at the Philips Research Laboratories (2). The Bartels monochromator (Fig. 5) uses a series of four Bragg reflections from a set of perfect germanium crystals to monochromate and collimate the X-ray beam. The beam that emerges is close to synchrotron quality in its monochromation and collimation. Recent developments using multilayer Xray mirrors even have intensities approaching those of a synchrotron (5 ). A thin-film diffractometer for epitaxial structures has some unique properties—in particular, the need for “high resolution”. The diffraction peaks are extremely narrow, often only a few seconds of arc. The diffractometer must step and scan extremely precisely. In fact the step size must be of the order of 0.0001°, with corresponding accuracy and precision. There are several solutions to this. The simplest is to use a stepper motor drive with reduction gearbox. However, there are costs to be paid in speed and mechanical tolerances. Modern high-resolution diffractometers are generally driven by synchronous motors with direct positional feedback from shaft-mounted optical encoders. X-ray Reflectivity
Figure 6. X-ray reflectivity curve.
silicon and germanium) on a substrate of Si. The sharp peak on the right is from the Si substrate. The broader, weaker peak on the left is from the thin layer. Note that the peak widths and separations are measured on an arc second scale. The entire scan range shown above is less than 0.2°—a typical peak width in a powder diffraction measurement. Note also that this measurement is shown on a logarithmic intensity scale covering more than 5 orders of magnitude. The composition of the film (%Ge) is obtained from the peak separations and the layer thickness is obtained from the width of the layer peak and the period of the low-intensity oscillations (thickness fringes). To deal with epitaxial thin films, the incident X-ray beam has to be very well defined both in terms of direction and in terms of wavelength. In fact, the requirements are extremely stringent indeed. The beam must be parallel to within better than a few thousandths of a degree and, to exploit the full power of the technique, must be more monochromatic than the natural width of the CuK α1 emission line itself! Any deviation
There is a third X-ray scattering technique that can be applied to thin films: X-ray reflectivity (6 ). The refractive index of solid materials in the X-ray wavelength region is slightly less than one. As an X-ray beam passes through the air–solid interface it is refracted to a slightly lower angle. The angle of refraction is small, a few tenths of a degree. As the incidence angle is reduced toward zero we reach the critical angle, at which the beam is totally reflected. Above the critical angle, the bulk of the beam is refracted into the sample and a small portion is specularly reflected. If the sample is a thin layer, the specularly reflected beams from the upper and lower surfaces of the layer interfere as the incidence angle is varied. The resulting pattern, shown in Figure 6, is known as an X-ray reflectivity curve. The example shows a full reflectivity curve from a sample consisting of two metal layers on a flat substrate. The highfrequency fringes contain information on the total layer thickness; the beating effect is due to the thinner of the two layers. X-ray reflectivity is a very interesting technique. It relies only on the density difference between layer and substrate, so it can be applied to amorphous as well as crystalline thin films. For many years it was regarded as a research curiosity. However, in the past decade X-ray reflectivity has been recognized as an absolute film-thickness measurement method. It is now certain to become one of the key metrology techniques in the next generation of semiconductor fabs (7). The optical configuration used for X-ray reflectivity is shown in Figure 7. As in thin-film XRD, a highly collimated X-ray beam is incident on the sample. Typically, the beam is parallel to within a few hundredths of a degree. The angle of incidence varies from 0° to approximately 2° during the course of the measurement. The reflected beam is collimated to a similar degree.
JChemEd.chem.wisc.edu • Vol. 78 No. 5 May 2001 • Journal of Chemical Education
615
Waters Symposium: X-ray Diffraction of Powders and Thin Films Table 1. Key Requirements of X-ray Diffraction Techniques Technique
Collimation of Incident Beam (deg)
Sample Motions
Collimation of Diffracted Beam
Bragg–Brentano
Few 1/10th's
Sample surface must be very flat and precisely aligned with goniometer axis; optional spinning motion
Very precisely positioned receiving slit
Thin-film XRD
Few 1/100th's
Precise height adjustment
Parallel plate collimator
Reflectivity
Few 1/100th's
Very precise height adjustment
Very precisely positioned receiving slits
High-resolution XRD
Few 1/1000th's
Very precise sample axis; tilt, rotate, height X-Y (for mapping wafers)
Relatively large aperture
At this point, it is interesting to summarize some of the features that distinguish the different thin-film diffraction instruments described above (Table 1). The Convergence By the late 1980s it was possible to buy a powder diffractometer, a thin-film diffractometer, a high-resolution diffractometer, or an X-ray reflectometer from any of the major X-ray equipment manufacturers. At that time, there was a widely held belief that the requirements of the various techniques were so diverse that they were best served by dedicated diffraction instruments. At least one X-ray diffraction manufacturer (Bede Scientific) focused exclusively on one application, thin epitaxial films. At Philips Analytical the entire product philosophy was based on “factory aligned–dedicated systems”. This market model neatly segmented customers, their applications, and X-ray diffraction instruments into well-defined categories. The plan was to deliver factory-aligned “solutions” to the customer. Unfortunately, thin films didn’t fit into the model. High-temperature superconductors, which appeared on the scene around 1990, illustrate the problem very nicely. The HTc superconductors such as YBaCuO are ceramic materials that were first synthesized as powders. Almost immediately, work began to grow high-quality epitaxial films of these materials using single-crystal substrates of the perovskite family, notably SrTiO3. Laboratories engaged in this work were dealing with the full range of materials—from powders through randomly oriented polycrystalline thin films to highly oriented polycrystalline films and ultimately to high-quality epitaxial films (8). Not only that. It wasn’t unusual for one sample to consist of a single crystal component (the substrate), an “epitaxial” component, and a randomly oriented polycrystalline component. The need arose for an X-ray diffraction instrument that transcended the boundaries of the conventional instrument packages available at that time. The solution was an X-ray diffractometer combining the attributes of all of the above diffractometers—and the ability to transform between applications without need for time-consuming realignment. The X-ray optics form the most critical part of the “transformation”. The collimation and monochromation devices in the incident and diffracted X-ray beams define the resolution of the diffraction measurement. These are also the most critical components as far as alignment is concerned. A number of solutions have been implemented to facilitate the exchange of X-ray optics. At Philips, we converged on a solution using prealigned optical modules on a system of kinematic mounts. The diffractometer becomes, in essence, an X-ray optical bench and the optical components are selected to achieve the resolution needed for a particular kind of sample or type of measurement. In other words, the sample characteristics and what you want to know about the sample dictate the optical configuration of the diffractometer. Figure 8 shows the incident616
Figure 7. Schematic of apparatus for X-ray reflectivity.
X-ray mirror Tube tower on PREFIX unit prefix interface (small) B Ge [220] Monochromator (large crystals)
A 2-positions PREFIX interface (large)
Figure 8. Kinematic mounting system for X-ray optical components.
beam mounting platform, with two optical modules and the X-ray source. A similar platform is constructed on the diffracted beam side. Using this concept, it has been possible to create a family of optical modules for all possible applications. The concept has been so successful that all modern X-ray diffractometers are now, to some extent, equipped with an optical interchange capability. Literature Cited 1. Jenkins, R. J. Chem. Educ. 2001, 78, 601–606. 2. Bartels W. J. J. Vac. Sci. Technol., B 1983, 1, 338–345. 3. Jenkins, R.; Snyder, R. L. Introduction to X-Ray Powder Diffractometry; Wiley: New York, 1996. 4. Bowen, D. K.; Tanner, B. K. High Resolution X-Ray Diffraction and Topography; Taylor and Francis: London, 1998. 5. Schuster, M.; Gobel, H.; Siemens, A. G. J. Phys., D 1995, 28, A270–A275. 6. Parrat, L. G. Phys Rev. 1954, 95, 359. 7. Stommer, R.; Gobel, H.; Martin, A. R.; Hub, W.; Pietsch, U. Semiconductor Int. 1998, 21 (5), 81–88. 8. Houtman, E.; Ryan, T. W.; David, B.; Dormann, V. Adv. X-Ray Anal. 1992, 35, 205–210.
Journal of Chemical Education • Vol. 78 No. 5 May 2001 • JChemEd.chem.wisc.edu
