Using Cathodoluminescence for the Inspection, Metrology and Failure Analysis of LED Processing

A significant amount of semiconductor devices being produced at present include light-emitting diodes (LEDs) which are preferred for the majority of lighting applications, including room lights and display technologies such as high definition televisions, augmented reality headsets and smartphones.

LEDs have an approximately 3x longer lifetime and a 10% higher efficiency compared to compact fluorescent light bulbs. Additionally, the devices are efficient, directly addressable and small.

Display devices employing µLEDs, including (left) LED television, (middle) augmented reality head-mounted display, and (right) smartphones.

Figure 1. Display devices employing μLEDs, including (left) LED television, (middle) augmented reality head-mounted display, and (right) smartphones. Image Credit: EDAX

The process of manufacturing μLEDs and different semiconductor devices can include several hundred stages and can have a duration of several weeks.

It is essential to determine and analyze the viability of devices in the fabrication stage to avoid the additional processing of defective devices. Failure analysis (FA) is also crucial to improve the throughput of future batches and to enhance processing stages. Figure 2 presents a simplified workflow for wafer-mounted μLEDs.

Flow chart illustrating device production for µLEDs presented.

Figure 2. Flow chart illustrating device production for μLEDs presented. Image Credit: EDAX

While there is a demand to increase the yield of μLED devices, non-destructive techniques for the identification of defects are not common, particularly those that can be employed in between production stages to determine handling defects, electrical shorts, semiconductor composition metrology, or dielectric layer faults.

The release of the Gatan Monarc® cathodoluminescence (CL) detector for scanning electron microscopes (SEMs) aims to solve these challenges by providing the characterization of luminescence from 6" wafer level to the nanoscale.

The Monarc detector additionally offers the most precise correlation with different signals. This feature enables compositional, crystallographic and topographical data to be acquired at once, which significantly enhances failure analysis capabilities.

Methods

CL, the light generated from a sample excited with energetic electrons, has proven to be highly beneficial for the compound semiconductor industry in device characterization and process development and is an effective tool for the analysis of dielectric materials.

This spectroscopic analysis enables materials to be identified that are not easily characterized by traditional SEM or optical analysis1. A main benefit of CL carried out in an electron microscope is that the nanometer-sized analysis spot can be employed for full-wafer inspection (and metrology) to the nanoscale characterization of specific micro-LEDs.

The Monarc CL detector is attached to a scanning electron microscope.

Figure 3. The Monarc CL detector is attached to a scanning electron microscope. Image Credit: EDAX

In analytical sciences, energy dispersive spectroscopy (EDS) in the SEM is a primary method for identifying the elemental composition with microscopic spatial resolution.

Utilizing the EDS technique in failure analysis (FA) is effective as it delivers analyses of the energy distribution of x-rays generated from a sample, from which the elemental distribution can be identified.

A CL spectrum captured from rare-earth-doped ceramic, revealing emissions across the ultraviolet, visible, and near-infrared portions of the electromagnetic spectrum.

Figure 4. A CL spectrum captured from rare-earth-doped ceramic, revealing emissions across the ultraviolet, visible and near-infrared portions of the electromagnetic spectrum. Image Credit: EDAX

Results and Discussion

In this investigation, commercially manufactured GaN-InxGa1-xN multiple quanta well (MQW) based mini-LEDs fabricated on a 4-inch sapphire substrate were analyzed utilizing the EDAX Octane Elite EDS and Gatan Monarc CL systems, both secured to an FE-SEM.

As shown in Figure 5, each rectangular LED was made from a stack of materials grown and selectively etched in series. From the bottom up, these included a thin AlN buffer layer, ~2 μm of n-GaN, ~2 μm of Si-doped N+-GaN, 200 nm of MQW layers, 300 nm of Mg-doped P-GaN, 100 nm of indium tin oxide (ITO), coated with ~300 nm of SiO2 and metal contacts.

(left) Low magnification view of a partially completed LED wafer. (middle) Secondary electron image of a single LED. (right) Cross-section schematic of InGaN/GaN MQW LED.

Figure 5. (left) Low magnification view of a partially completed LED wafer. (middle) Secondary electron image of a single LED. (right) Cross-section schematic of InGaN/GaN MQW LED. Image Credit: EDAX

Although the SEM image is useful for the presentation of surface morphology, data is lacking in regard to the consistency of LED material defects and distribution. LED devices were analyzed using CL imaging and multiple defects used for FA were revealed to assess material composition, handling-induced damage and fabrication defects.

Handling-Induced Defect

Unfiltered CL and secondary electron images were acquired across a large region (~16 mm2) (Figure 6). The CL image demonstrated that approximately 1% of LEDs generate low brightness, a defect that was not obvious from the equivalent SEM images.

The devices that were affected were in an 'X' pattern that did not relate to the row pattern and device column and were likely a result of mishandling during the process of fabrication. These defective devices were analyzed to establish the source of the defect and whether the devices are viable for use.

(left) Secondary electron and (right) CL image of mini-LED array. CL image reveals a series of defected LEDs in the shape of an ‘x’ with reduced intensity.

Figure 6. (left) Secondary electron and (right) CL image of mini-LED array. CL image reveals a series of defected LEDs in the shape of an ‘x’ with reduced intensity. Image Credit: EDAX

Presented in Figure 7, a CL spectrum image (hyperspectral map) was collected for one of the compromised LEDs (yellow box). The CL maps indicate a surface influence mainly influencing the emission of the MQW layer, a mechanical scratch, produced after the MQW but prior to metal deposition.

(left) Secondary electron image, (center) CL spectrum image displaying wavelengths from 300 – 700 nm, and (right) colorized spectrum image highlighting emission from GaN (360 nm, green colorization), and InxGa1-xN (430 nm, blue colorization; and 460 nm, red colorization) with 20 nm bandwidth. Each color band is normalized by intensity.

Figure 7. (left) Secondary electron image, (center) CL spectrum image displaying wavelengths from 300 – 700 nm, and (right) colorized spectrum image highlighting emission from GaN (360 nm, green colorization), and InxGa1-xN (430 nm, blue colorization; and 460 nm, red colorization) with 20 nm bandwidth. Each color band is normalized by intensity. Image Credit: EDAX

The electrical pathways are likely to have been impeded by these defects which would greatly circumvent or decrease the emission of LED light, and they should be excluded from additional processing.

CL enables the user to establish which devices are defected and produces a map that can be employed to selectively reject compromised devices from additional processing.

Defects like these can result in the exclusion of significant regions of wafer area, which support dozens of devices, and are often the most avoidable as they emerge from human interactions, for example during the exchange between instruments in processing or wafer transport.

Electrical Defect

Defects that impact the electrical connections in LEDs can affect an entire array and results in the degradation of packaged device performance. The most popular electrical defects are short circuit and open circuit.

(left) Secondary electron and (right) unfiltered CL images of LEDs with missing metal contact pads.

Figure 8. (left) Secondary electron and (right) unfiltered CL images of LEDs with missing metal contact pads. Image Credit: EDAX

LEDs were found to have missing metal contact pads and are depicted in Figure 8, which results in a device that can function as an open circuit electrically.

The bright areas in the CL image where the metal layers should have been allowed the missing contact pads to be easily identified. The CL brightness was 2.5x brighter compared to areas with cladding due to the absence of the cladding layers.

(left) Secondary electron image and (right) corresponding CL image of an electrical short defected LED.

Figure 9. (left) Secondary electron image and (right) corresponding CL image of an electrical short defected LED. Image Credit: EDAX

Shown in Figure 9, an LED with an electrical short defect was identified. These types of defects are critical as they can lead to power loss and circumvent the functionality of LEDs.

The shadow produced by the contacting material shows a bright to dark contrast ratio of up to 7:1 with the area surrounding it, while the contrast ratio is only 2:1 for the secondary electron image.

The observation of CL images greatly simplifies the discovery of electrical short defects. The enhanced contrast delivered by the CL map allows more robust automatic failure detection algorithms to be created with a subsequent increase in the yield of devices.

Fabrication Defects

(left) Unfiltered CL image and EDS material composition results of a defect LED. EDS material mapping for gallium (red), silicon (yellow), and oxygen (green).

Figure 10. (left) Unfiltered CL image and EDS material composition results of a defect LED. EDS material mapping for gallium (red), silicon (yellow) and oxygen (green). Image Credit: EDAX

As shown in Figure 10, CL mapping additionally uncovered a number of other defects, such as fabrication defects close to the upper and lower device contacts. In this example, CL did not offer a clear explanation for the nature of the defect. CL imaging and EDS mapping were carried out at the same time to establish the defect type.

Presented in Figure 10, the EDS maps uncovered local deficiencies in oxygen and silicon and an excess in gallium for the defects, represented by the red box produced from SiO2 being excluded by accident during one of the liftoff processing stages. An excess of SiO2 was revealed by other areas (green box).

Material Composition

(left) CL spectrum image displaying wavelengths from 300 to 700 nm, (center) colorized spectrum image highlighting emission from GaN (360 nm, green colorization), and InxGa1-xN (430 nm, blue colorization; and 460 nm, red colorization) with 20 nm bandwidth. Each color band is normalized by intensity. (right) LED CL spectra from (red) MQW region and (blue) central GaN region (increased by a factor of 10 for clarity).

Figure 11. (left) CL spectrum image displaying wavelengths from 300 to 700 nm, (center) colorized spectrum image highlighting emission from GaN (360 nm, green colorization), and InxGa1-xN (430 nm, blue colorization; and 460 nm, red colorization) with 20 nm bandwidth. Each color band is normalized by intensity. (right) LED CL spectra from (red) MQW region and (blue) central GaN region (increased by a factor of 10 for clarity). Image Credit: EDAX

To the left of Figure 11, the CL image depicted a very high emission intensity suggesting some changes across the LED surface, which was not concealed by opaque metal layers.

The center of Figure 11 presents a spectrum image that was acquired and colored to show differences in the indium-fraction distribution of the LED MQW. Figure 11 (center) consists of three bandpass images that were taken from the spectrum image, colorized and stacked together.

Emission at 430 nm is represented by the blue intensity, while the red originates from 460 nm and the green at 360 nm, each with a bandwidth of 20 nm.

This composite bandpass image uncovers areas where the red intensity is decreased, suggesting lower indium concentrations locally by around 0.4% less than the 16.2% (In) average. This variance is not large enough for the device to be excluded from additional processing.

Spectra taken from the spectrum image, blue (x10) from the GaN region and red from the MQW, are depicted in Figure 11 (right) and showed a very high intensity close to 450 nm, which corresponds to the InGaN emission and the GaN at 365 nm.

Summary and Conclusions

The manufacture of semiconductor devices requires an extensive amount of resources and time. As such, inspection and metrology are crucial to ensure that the allocation of resources is not spent on defective devices.

CL with supplementary EDS delivers essential inspection methods for the direct observation of various defect types which can result in a significant decline in the performance of the device.

CL also has the capability to image layers underneath the surface, assists in the development of more robust defect identification algorithms based on enhanced contrast, and allows defect impact to be determined through the measurement of device intensity.

EDS offers the ability to differentiate types of defects by determining the elemental composition of the defected areas.

References

  1. D. J. Stowe, J. D. Lee and M. Bertilson, "Octane Elite and Monarc Come Together to Capture EDS and CL Simultaneously," EDAX Insight, pp. 1-2, September 2020. 

This information has been sourced, reviewed and adapted from materials provided by EDAX.

For more information on this source, please visit EDAX.

Citations

Please use one of the following formats to cite this article in your essay, paper or report:

  • APA

    EDAX. (2021, March 16). Using Cathodoluminescence for the Inspection, Metrology and Failure Analysis of LED Processing. AZoM. Retrieved on October 16, 2021 from https://www.azom.com/article.aspx?ArticleID=20146.

  • MLA

    EDAX. "Using Cathodoluminescence for the Inspection, Metrology and Failure Analysis of LED Processing". AZoM. 16 October 2021. <https://www.azom.com/article.aspx?ArticleID=20146>.

  • Chicago

    EDAX. "Using Cathodoluminescence for the Inspection, Metrology and Failure Analysis of LED Processing". AZoM. https://www.azom.com/article.aspx?ArticleID=20146. (accessed October 16, 2021).

  • Harvard

    EDAX. 2021. Using Cathodoluminescence for the Inspection, Metrology and Failure Analysis of LED Processing. AZoM, viewed 16 October 2021, https://www.azom.com/article.aspx?ArticleID=20146.

Ask A Question

Do you have a question you'd like to ask regarding this article?

Leave your feedback
Submit