Wednesday, October 17, 2007Volume 7, Issue 5
Equipment Update
EOC Extends Characterization Capabilities for Advanced Wide Band-Gap Materials and Devices
Joshua A. Robinson, Research Associate, EOC Materials Division
Semiconductor device performance is largely based on the ability to produce and tailor ultra-pure starting materials. Silicon carbide (SiC) and related semiconductors exhibit material properties that are advantageous for high power, high temperature applications. Their wide band-gap, thermal conductivity, high electron-hole pair creation and displacement energies are all necessary for long-term operation in harsh environments that may include high temperatures, intense radiation, or continuous high voltage electrical stressing. However, current SiC material quality has limited its wide spread implementation. In particular, crystal defects are considered to be a major factor in blocking large area devices used in today's high power technology. Micropipes, screw dislocations, and morphological defects are just some of the material properties that reduce crystalline quality and decrease carrier lifetime in a semiconductor material, which means theoretical performance predictions may not be achieved.
The Penn State Electro-Optics Center was created in 1999 to serve the needs of the Department of Defense. This includes research and development specifically in the areas of fiber optics, sensor technology, laser development, device reliability, and materials design and processing. The Materials Division of the Electro-Optics Center has a successful history in the development of growth techniques for wide band-gap materials such as silicon carbide, nitride semiconductors, complex oxides, and novel crystal materials such as lutetium oxyorthosilicate (LSO). As part of our efforts we have focused on expanding the characterization techniques currently available within the research community of the Materials Research Institute to include characterization of advanced wide band-gap materials and devices. In the following article two characterization techniques are described that yield critical information for improving material properties and device design.
X-ray Diffraction: Wafer Scale Mapping of Semiconductor Crystal Quality
Figure 1: MATSE Student Melisa Steighner places a
2-inch SiC wafer, grown at the Penn State EOC, in the
PANalytical X-Pert PRO MRD x-ray diffraction system
located at the Material Research Institute in University Park.
X-ray diffraction is a means to identify the quality of the starting crystal in a non-contact, non-destructive manner. Information on the strain in a crystal can be delineated by observing shifts, broadening, or both, in a diffraction peak. The Penn State Materials Characterization Lab (MCL) offers several x-ray diffraction techniques such as glancing angle diffraction, Theta - 2 Theta (Figure 1), and rocking curve x-ray diffraction to identify crystal structure, quality, grain size, or if the material is indeed single crystal. X-ray wafer mapping is a means to measure both peak shift and peak width as a function of position on wafers up to 3 inches in diameter. To date, however, utilization of x-ray diffraction wafer mapping to identify crystal quality and correlate that with device performance has not been explored at Penn State.
The Materials Design and Process Technology Division at the PSU-EOC has recently begun implementing x-ray diffraction wafer mapping to identify non-uniformities in wafers purchased from industry and grown within the division. Our focus has been on mapping SiC substrate wafers and correlating crystal quality with the quality of epitaxial layers (hetero- and homo-epitaxial) grown on these substrates. Figures 2 and 3 are x-ray wafer maps of a 2-inch diameter SiC substrate. Wafer mapping consists of performing x-ray rocking curves at given x,y positions on a wafer that is to be used in subsequent experiments. The goniometer arm is placed at the theoretical Bragg angle (75.122 - for SiC (008)) and the sample is subsequently "rocked" to either side of the incident angle that satisfies the Bragg condition. If the crystal is perfect, the Bragg condition will only be satisfied at exactly - the goniometer arm position. This is rarely the case and what actually happens is that there is a finite width where the crystal will satisfy the Bragg condition. Some rocking curve width is the result of the x-ray optics used, but the majority of a width measured in an x-ray rocking curve is a result of the crystalline quality. The crystal quality is extracted by observing how wide a sample can be rocked and still meet Bragg conditions. Samples that can be rocked farther and still diffract are considered to be of lower quality.
Figure 2: X-ray diffraction rocking curve wafer map indicating
peak width as a function of position on a 2-inch SiC wafer
substrate. Large peak widths indicate regions of lower
quality crystal.
Figure 3: X-ray diffraction wafer map indicating the 2? peak
position as a function of position on a 2-inch SiC wafer substrate.
Note that regions with shifted peaks are correlated closely with
peak widths in Fig. 2.
Figure 2 is a map of the (0 0 12) SiC peak width, which indicates the level of non-uniform stress in the material. Regions within the wafer that exhibit higher levels of stress (lower quality) yield wider peak widths than those regions of little or no stress. Figure 3 is a map of peak position, a measure of uniform stress in a material, of the same SiC substrate wafer. Shifts towards lower angles means the atoms within the crystal are under tensile stress. SiC is known to exhibit wafer bow or curvature that results from residual stress in the bulk crystal. Curvature in the wafer may be visualized through x-ray wafer mapping by observing a gradual shift in peak position, similar to that in Figure 3. It can also be seen that changes in the peak width in Fig. 2 correlate well with the non-uniform peak shifts in Fig. 3.
Silicon carbide is set to replace Si-based high power electronics, and is primed for many other applications in the near future. The ability to identify crystal variations across entire wafers is an important aspect in the understanding and engineering of state-of-the-art devices and detectors. Using x-ray diffraction instrumentation at Penn State, we have mapped wafers as a means of quality control for the subsequent growth of high quality epitaxial layers, and as a first step towards understanding how substrate quality affects device performance.
Electron Beam Induced Current
Electron beam induced current (EBIC) is one of the most recent characterization techniques being implemented at Penn State to assist in correlating device performance with material quality. The EBIC method is widely used for the identification of localized defects and carrier (electrons or holes) lifetimes in semiconductor materials. Figure 4 is a schematic setup of the EBIC process. EBIC relies on using an electron beam, generated in a scanning electron microscope (SEM), to generate electron-hole pairs in a semiconducting material. The electrons and holes, known as "free carriers," then drift through the material and are separated by an electric field that is generated by applying a voltage between two metal layers on either side of the sample. Subsequent to separation, they are collected and measured as an EBIC current. The current measured in an EBIC experiment is quite small, and as a result an amplifier is required to boost the signal so external measurement systems can collect and store the signal.
Figure 4: Schematic EBIC setup. Carriers are generated
by an electron beam in the semiconductor; they are then
separated by an electric field and subsequently collected.
Carrier lifetime is extracted by monitoring the collected current as a function of the distance the electron beam is away from the edge of the top metal film. Figure 5 illustrates EBIC current as a function of the distance between the metal edge and the beam as distance is increased. Similar to sound and light waves, free carriers are scattered by objects (impurities or defects) that disturb their motion between the source (electron beam) and detector (electric field and metal contacts). As the beam is moved away from the metal edge the number of defects or impurities that an electron will encounter will increase and thus more scattering will occur, which results in a decrease in the number of collected carriers and hence a reduction in the EBIC current. By plotting the current versus distance and measuring the slope of the curve, we are able to identify the distance a carrier can travel before being annihilated. Carrier lifetime is a direct indication of material quality and device performance - the farther an electron can travel before encountering a defect, the better the device should perform.
Figure 5: Illustration of the measured EBIC current
as a function between the electron beam and top metal edge.
Additionally, using the imaging capabilities of the SEM, we are able to image defects and correlate device performance with defect type and density. As shown in Figure 6, defects within a device will show up as either bright or dark spots in the image. The type of defect can be correlated with shape, size, and contrast in the EBIC image. Qualitatively, the brighter images indicate longer carrier lifetimes and thus higher material quality. High power devices often mimic Figure 6a where defects in the device show up as dark spots, indicating that there are regions of defective material acting as a carrier recombination center. Devices that include regions of intentionally introduced impurities will look similar to Figure 6b. Uniform introduction of impurities will decrease the carrier lifetime except in those regions where the ion implantation did not degrade the crystalline quality - these are indicated by the bright spots.
Figure 6: EBIC images of (a) Schottky Diode
and (b) P/N Junctions in SiC.
Improving our understanding of the defects present in a material means we are able to tailor the material to reduce unwanted defects and ultimately improve the device performance and reliability. EBIC is one way we are able to directly image defects and measure carrier lifetimes in devices, and thanks to the wide range of complementary techniques available within the Penn State MCL, we are able to take the EBIC technique one step further. Utilizing the combination of EBIC and the Focused Ion Beam (FIB) (see Figure 7) we are able to identify the location of defects and "carve" out the area of interest for further investigation on the atomic level in a transmission electron microscope. Integration of EBIC and the FIB is a powerful technique and research within the full spectrum of the Penn State materials community will benefit.
Advancing knowledge through the characterization of materials has led to a rich history of high quality materials synthesis, device design, and device fabrication at Penn State. X-ray wafer mapping yields crystal information on the large scale with the ability to map wafers up to 3 inches in diameter. EBIC, on the other hand, has the ability to identify defects on the microscale and yield vital information on carrier lifetime and defect densities within device architectures. Combining these techniques with those previously established in the MCL, we have the ability to intimately characterize every step from material synthesis to device fabrication and operation.
Figure 7: Photograph of the Focused Ion Beam Instrument,
which can be combined with EBIC for site specific
TEM sample preparation of defect regions.
These techniques are available through the Materials Characterization Laboratory and can be accessed by all those interested in materials research. Please contact Mark Angelone (Validate to view address - Send Email via form) for instruction on x-ray wafer mapping and EBIC. Contact Trevor Clark (Validate to view address - Send Email via form) for instruction on the FIB.
Joshua A. Robinson earned his Ph.D. at Penn State studying the material properties of metal/semiconductor interfaces and the influence of processing conditions on metal/semiconductor contact performance. After a short tenure as a post-doctoral scholar at the Naval Research Laboratory researching high sensitivity carbon nanotube sensors, Dr. Robinson returned to PSU to join the Materials Division of the Electro-Optics Center. His current research interests include wide bandgap semiconductor materials for power electronics, radiation detection, and high temperature sensors. You can reach Josh Robinson at (Validate to view address - Send Email via form).
