Triple Axis Analyzer and Reciprocal Space Mapping

When measuring Omega-2Theta scans in double axis configuration, there is still a component in the scan due to scattering within the plane, as a simple slit allows this scatter into the detector . This is reduced as the slit is reduced. However, as the size of the slit is reduced the intensity is also reduced. To reduce this to acceptable acceptance angles, often a slit of 0.1mm is required, reducing the intensity by up to 95%. Clearly this is not the best solution to obtain a high resolution at the detector.

Above is a double axis Omega-2Theta scan of a GaAs substrate with a graded InxGa1-xAs buffer layer. The tallest peak is the substrate peak, while the hump trailing off on the left is the buffer layer. The sharp peak on the right was a surprise. In this case, we did not know whether this peak was due to an additional layer with a different lattice parameter, a relaxed layer or from a tilted portion of the material.

The chart above is the triple axis map, plotted in diffractometer coordinates. The horizontal axis represents d-spacing changes (w-2q), while the vertical axis is mosaic tilt (w). The tall peak at (0,0) (in red) is the substrate peak. There is also another tall peak that has the same d-spacing as the substrate, but is shifted in the omega direction. This was the sharp peak that appeared on the right in the double-axis scan. From the triple axis map, we can determine that the sharp peak on the right is due to a tilted portion of the substrate, since the d-spacing matches exactly that of the substrate peak, while the peak is from a portion of material tilted with respect to the bulk of the substrate. The layer to the left of the substrate peak also changes in w as well as w-2q. This indicates the compositionally graded layer is also tilted, with the epilayer tilt increasing with misfit.

Direct Relaxation Measurement

A key requirement in both compound and Si semiconductor analysis is the determination of the relaxation of individual layers within a multiple layer stack. The strain and composition of the layers are critical, as this sets the performance criteria. The Jordan Valley method, developed by Bede, is a unique method which has been adopted as the standard process for monitoring relaxation in production lines.

Traditional scanning methods

• Low, broad peaks for thin layers leading to inaccurate relaxation and composition determination
• Relaxation value needs to be assumed to obtain meaningful data
• Cannot separate individual layers from a multilayer stack
• Slow measurements
• Manual interpretation required

The method developed by Bede, and now used by Jordan Valley, determines the layer relaxation without assuming anything about the structure. It uses the knowledge of the out of plane lattice parameter (from the (004) symmetric scan) to directly probe the relaxation values by scanning across relaxation in reciprocal space. With this unique method, different layers with differing out of plane lattice parameters can be probed for either relaxation or composition without the influence of other peaks obscuring the peaks of interest.

The above example is the unambiguous automated measurement of the composition and relaxation of a SiGe layer on Si. The (004) scan determines the exact start and end points of the relaxation scan in reciprocal space. As can be seen from the relaxation scan, the value for the relaxation is read directly from the peak position, with no complex interpretation required. In this example the layer is fully strained with a composition of ~ 13% Ge.

High Resolution X-ray Diffraction (HRXRD)

High resolution X-ray diffraction has long been used in the compound semiconductor industry for the characterization of epitaxial layers. Traditionally it has been used in the determination of thickness and composition of the epi-layers, but more recently the technique has advanced to enable the determination of strain and relaxation within a given layer of a multilayer structure. The example data below shows a measured HRXRD scan – symmetric 004 reflection from a single layer on a bulk substrate.

Scan is typically done by scanning sample and detector in 1:2 ratio. The substrate peak is normally the sharpest and most intense feature in the scan. The position of the Bragg peak is determined by Bragg's law, i.e. the lattice parameter of the substrate, the wavelength of the X-rays. The layer peak that in this example is on the left hand side of the substrate peak is visible. The lattice parameter of layer is larger than that of substrate in this example, as from Bragg's law it diffracts at smaller angles than the substrate. The difference in peak position relates to the difference in lattice parameter. These can be related to composition, strain or relaxation of the layer. On either side of the layer peak there are interference fringes that result from interference of the wavefields in the layer. These features can be used to accurately determine the layer thickness.

Once the data have been collected, the analysis of the data is required to extract the parameters of interest. The most common method used by Jordan Valley customers, and widely used by other HRXRD experts, is using the JV-HRXRD analysis software (formerly Bede RADS). This software can either be used as off-line software by the user, or automatically as part of the programmed measurement cycle on the tool itself. An example of a GaAs HEMT structure is shown below.

Jordan Valley introduces a Comprehensive Process Control solution for LED / PV / Epi-layer wafer manufacturers.

This consists of:

• High productivity QC-Velox / QC3
  • Specifically designed HRXRD for compound epi layers
  • Optimised for production
• Analytical Software (RADS)
  • Fast and accurate calculation of process parameters
• Results and Data reporting
  • Fast feedback to production

Implementing this comprehensive solution will enhance your yeild and increase profitability.