1 Introduction

There is extensive research focused on improving the efficiency of LEDs [1, 2]. While regular LEDs are commonly used in display technologies, there is a growing trend towards employing micro-LEDs to enhance display characteristics [3,4,5,6,7]. Robust display technologies such as liquid crystal display (LCD) and organic light-emitting diode (OLED) exist, but the industry is shifting towards displays with micro-LEDs [8, 9]. The main focus in display technology is on displays populated with three-color micro-LEDs due to their superior efficiency and light characteristics [10, 11].

Micro-LED displays would benefit from fast LED response as well as energy savings. However, as micro-LEDs become smaller, there is an increase in surface recombination areas [12], specifically the sidewalls relative to the emitting areas. This can lead to lower quantum efficiencies compared to larger LEDs. In addition, the radiative recombination of micro-LEDs on the sidewalls can affect the output light emission angle. To ensure the successful mass transfer of micro-LEDs to displays, it is crucial to perform quick quality control during the fabrication of micro-LEDs and/or the display itself.

During the regular mass transfer process, micro-LEDs are fabricated and transferred to the display without being tested. The backend processes for building the display must be completed before measuring and testing the display. However, any defect in a micro-LED or the display can result in a faulty display, with each malfunctioning micro-LED leading to a non-functioning pixel. Therefore, achieving a faultless fabrication process is highly challenging. To address this issue, the use of a micro-photoluminescence microscope during the fabrication processes can help identify and remove defective micro-LEDs or repair faulty display pixels before the backend processes. Early fault recognition in such an expensive process path is highly desirable.

In this article, we present the development of a micro-photoluminescence microscope and demonstrate its effectiveness in distinguishing malfunctioning micro-LEDs early in the display fabrication process, thereby saving valuable resources. We also utilize the microscope to conduct tests that show a significant reduction in micro-photoluminescence close to the sidewalls of the micro-LEDs.

2 Experimental setup and results

In the past, researchers have used micro-photoluminescence for investigating micro-LEDs [13,14,15,16]. However, we have specifically built our own micro-photoluminescence microscope and utilized it during the mass manufacturing of displays. Our setup, as shown in Fig. 1a, consists of a core microscope (a regular Nikon microscope) with lasers arranged around the head. These lasers emit light at a higher photon energy of 405 nm, which photo-excites the objects of interest, in this case, green micro-LEDs. The micro-LEDs absorb this light and re-emit it at the photoluminescence wavelength of 525 nm. The objective camera path is equipped with a filter and a camera, although a spectrometer can be used instead to measure the wavelength spectrum of the emitting micro-LEDs.

Fig. 1
figure 1

a Side view of the micro-photoluminescence microscope: the microscope consists of a microscopic lens with 405 nm lasers arranged around its head. During the photoluminescence mode, the optical microscope light is turned off, and the lasers become the sole sources of illumination. The green micro-LEDs are excited by the lasers, causing them to photoluminesce at 525 nm. The laser beams are represented by thin-dashed lines, and the photoluminescence beam is shown as a thick dash-dot line. b Top view of the photoluminescence microscope: In this view, seven lasers are arranged all around the head of the microscope to ensure uniform illumination. The micro-LED is depicted as a rectangle at the center of the laser beams, and the camera path is shown as a circle

Full size image

In Fig. 1b, the top view schematic of the micro-photoluminescence system is shown. Seven lasers are arranged around the sample to provide as uniform emission as possible. During the photoluminescence mode, the regular microscope light path is blocked, and only the regular optical path is used for navigating to the spot of interest.

The beam diameter of the 405 nm lasers used for photoexcitation is 1 mm. The laser beam is widened to 10 mm diameter on the target. The lasers are arranged with 60 degrees from the surface normal direction to the micro-LEDs. The laser intensities are fairly low and the beam is quite wide, so there is not too much problem with the laser speckles. The objective lens of the microscope for the current work is Nikon 20X. The optical filters used for photoluminescence are bandpass filters at 520 nm ± 10 nm. The black/white camera used for photoluminescence is 2000 × 2000 pixels and the pixel size is 18 μm. Green bandpass filters are used alongside with black/white camera to ensure that only green photoluminescence light is absorbed on the camera pixels. Note that we could have potentially used any illumination source and not just lasers as far as the photoexcitation is at shorter wavelength than the photoluminescence wavelength and with enough illumination power. A combination of the laser power for illumination and camera integration time is used to produce bright enough pictures for the purpose of the experiments. The camera integration time used in this work is 200 ms. The fluorescence and/or photoluminescence may occur in organic as well as inorganic materials such as LEDs [14]. The major requirement for proper photoluminescence is for the illumination wavelength to be of higher photon energy than the photoluminescence and in the case of micro-LEDs, the power to produce the light can be generated from the higher energy photons exciting the quantum wells of the micro-LEDs instead of the electrical excitation of the quantum wells. A micro-LED can be physically connected to the power source and the electrical excitation will result in micro-LED light or even without a physical electrical connection to the electrical source, a micro-LED quantum well may be excited by pumping in enough photon energy through micro-photoluminescence to produce the micro-LED light. The micro-LED light is far weaker than the photoexcitation power source since the conversion efficiency of the micro-photoluminescence is fairly low, and therefore, bandpass filters are used to cut the original excitation light and ensure that only excited micro-LED light can pass to the camera. The general view of the micro-LED is also shown in Fig. 2. The camera is also replaced with a spectrometer to ensure that the only light passing to the camera is the photo excited light from the micro-LED. Figure 3 shows the results from the spectrometer and confirms that the only light passing to the camera is the photoluminescence light from the micro-LED and at 525 nm. All other lights including the laser excitation light are cut by the bandpass filter.

Fig. 2
figure 2

General view of the micro-LEDs. The scale of the pictures is 10 μm

Full size image
Fig. 3
figure 3

All lights to the camera are cut by the bandpass filter and the only light which goes to the camera is the photoluminescence light from the micro-LED

Full size image

Figure 4 illustrates the micro-LEDs that have been mass transferred to a display. Without being tested with the micro-photoluminescence microscope, the display is turned on and found to be defective. Using regular microscope light and micro-electroluminescence, certain display pixels are not turning on (Fig. 4a and b), indicating potential faults in the micro-LEDs or the display processes. To narrow down the reasons for the malfunctioning display pixels, a micro-photoluminescence image is captured at the same spot. Figure 4c shows that certain micro-LEDs do not turn on during this mode, suggesting that faulty micro-LEDs are the most probable reason for some of the defective display pixels.

Fig. 4
figure 4

The images depict a three-color display, with only one color populated with micro-LEDs. a The display is connected to the electrical source and an optical image is captured. b The electroluminescence image is captured at the same location while the display remains connected to the electrical source. c The photoluminescence image is captured when the display is disconnected from the electrical source. From these populated displays, several issues can be observed. Some of the micro-LEDs, indicated by white ovals, do not turn on during photoluminescence, indicating malfunction before being placed on the display. This severe problem could have been detected earlier using a micro-photoluminescence microscope. The scale of the pictures is 10 μm

Full size image

A possible reason for a faulty micro-LED could be a defective quantum well and/or a short near the sidewall of the micro-LED. Surface defects on the sidewalls of micro-LEDs can significantly degrade their internal quantum efficiency [10, 17, 18]. Studies have found that the efficiency of micro-LEDs decreases with chip size due to sidewall trapping effects [19, 20]. Surface defects can generate nonradiative recombination and lead to attenuation of the device external quantum efficiency (EQE) [11]. The range affected by surface defects is approximately equal to the carrier diffusion length, typically 1–10 μm [11]. The presence of sidewall defects can also lead to the performance degradation of entire micro-LEDs, known as the “sidewall effect” [11]. Therefore, it is important to address and minimize surface defects and sidewall trapping effects in order to improve the performance of micro-LEDs.

In addition, there are certain display pixels that turn on during the micro-photoluminescence mode but not during the electroluminescence mode. These display faults can be associated with the backend display processes during encapsulation of the micro-LEDs and the connection of top contacts. Proper reprocessing of the micro-LED encapsulation can repair these types of faults. For example, a micro-LED quantum well might be photo excited and illuminate well since the excitation mechanism is through photo excitation rather than physical connection, as in the case for electroluminescence. On the other hand, electroluminescence depends on the proper connection of the micro-LED poles to the electrical source. Even though a photo excited micro-LED might show fine, an electrical connection to the power source might be faulty which may still result in dark micro-LEDs during electroluminescence. Therefore, each of the photoluminescence and electroluminescence modes of imaging is good for certain purposes. The photoluminescence mode can be employed to purely understand the health of the micro-LED, and the electroluminescence mode can be used to confirm proper connections.

By utilizing micro-photoluminescence during the manufacturing process, potential issues with the micro-LEDs and display can be identified earlier, allowing for more efficient troubleshooting and quality control.

Figure 4 illustrates the micro-LEDs that are transferred to a display without undergoing testing with the micro-photoluminescence microscope. However, when the display is turned on, it is found to be defective. Figure 4a and b, using a regular microscope light and micro-electroluminescence, shows that certain display pixels do not turn on. There could be several reasons for these faulty pixels, such as defective micro-LED quantum wells or defective display processes. To narrow down the possible reasons for malfunctioning display pixels, a micro-photoluminescence image is captured at the same spot. Figure 4c reveals that a certain number of micro-LEDs do not turn on during this mode, suggesting that faulty micro-LEDs are the most probable reason for some of the defective display pixels. One possible cause for a faulty micro-LED could be a defective quantum well or a short near the sidewall of the micro-LED. A possible short near the edge of a micro-LED can be captured by detailed failure analysis such as SEM as shown in Fig. 5. This problem could have been addressed earlier during the display process and/or micro-LED fabrication process.

Fig. 5
figure 5

SEM image of a micro-LED edge for possible shorts

Full size image

Figure 4c also demonstrates that there are certain display pixels that turn on during the micro-photoluminescence mode but do not turn on during the electroluminescence mode. These display faults can be attributed to the backend display processes during the encapsulation of the micro-LEDs and the connection of top contacts. Proper reprocessing of the micro-LED encapsulation can repair these types of faults.

In addition to micro-LED functioning, photoluminescence can also be used to characterize the brightness variations of the micro-LEDs. To achieve this, the brightness of the photoluminescence image is dimmed by reducing the camera integration time. Figure 6a displays the dimmed micro-photoluminescence of the micro-LEDs on the display. The bright lines next to the square micro-LEDs are due to the reflection of the micro-LEDs from the display mirrors, which are eliminated during data processing. The image reveals variations in brightness among the micro-LEDs, including variations within each micro-LED. The histogram of the photoluminescence brightness of the micro-LEDs is shown in Fig. 6b, with the brightness data normalized on the horizontal axis.

Fig. 6
figure 6

The scale of the pictures is 10 μm. a The micro-photoluminescence of the micro-LEDs on the display is intentionally dimmed to prevent saturation of the camera pixels. The tiny bright lines next to the micro-LEDs are reflections from the display mirrors. However, these mirror reflections are removed during data processing and do not directly impact the brightness results. b Histogram of the normalized photoluminescence brightness of the micro-LEDs

Full size image

Figure 7a represents a single micro-LED, with the graph indicating the vertically summed image pixel brightness over the micro-LED area. Figure 7b shows the same micro-LED, with the micro-photoluminescence signal summed horizontally over the micro-LED area. Figure 7 demonstrates that the micro-photoluminescence significantly drops near the sidewalls, as expected. The sidewalls are likely the main source of nonradiative recombination as the LED size decreases. For a square LED with an edge length of D and height of H, the ratio of the sidewalls to the central emitting area is 4H/D. Therefore, reducing the sidewall area compared to the general emitting area is desirable to improve the micro-LED quantum efficiency, which can be achieved by making the micro-LED as thin as possible. However, handling super-thin micro-LEDs mechanically poses significant challenges and may lead to a drop in yield.

Fig. 7
figure 7

Micro-photoluminescence shown for one micro-LED. The scale of the pictures is 10 μm. a The brightness of the image pixel is summed vertically and plotted on top of the micro-LED. This gives us insight into the overall brightness distribution. b Similarly, the brightness of the image pixel is summed horizontally. This helps us understand the contribution of different areas, particularly the sidewalls, to nonradiative recombination in micro-LEDs

Full size image

For further verification, all the micro-LEDs in Fig. 6 are cropped using Matlab image analysis toolbox, and the centers of the micro-LEDs are found and aligned. The sum of all vertical image pixel brightness for all micro-LEDs is aggregated and displayed in Fig. 8a. Figure 8b follows the same procedure but aggregates the sum of all horizontal image pixel brightness for all micro-LEDs. The brightness near the sidewalls of the micro-LEDs is observed to be around 20–40% of the brightness near the center of the micro-LEDs. Figure 8 clearly illustrates the significant drop in micro-photoluminescence signals near the sidewalls, indicating that the sidewalls are likely the major source of nonradiative recombination for micro-LEDs.

Fig. 8
figure 8

Aggregated micro-photoluminescence signal over all of the micro-LEDs. a The brightness of micro-photoluminescence is summed over the vertical image pixels for each micro-LED and aggregated over all of the micro-LEDs. b The brightness of micro-photoluminescence is summed over the horizontal image pixels for each micro-LED and aggregated over all of the micro-LEDs. The brightness near the edges is approximately 20–40% of the brightness near the center of the micro-LEDs. The approximate location for the edges of the micro-LEDs is displayed as dotted lines

Full size image

The article provides detection results for single images, but the whole display can be analyzed using the same method. A linearized movement on the microscope is used to capture the picture of the whole display. Therefore, the whole display can be analyzed using the same method as shown in the context.

3 Conclusion

A display without photoluminescence quality control is used for testing. After the backend processes are completed, the display is turned on and found to be defective. Optical and electroluminescence images are captured to identify the source of the pixel defects. A micro-photoluminescence microscope is built using a regular Nikon microscope, revealing that some micro-LEDs are faulty before being transferred to the display. Other display pixels are observed to turn on during photoluminescence mode but turn off during electroluminescence mode, indicating issues with the backend display processes. The brightness of the micro-LEDs’ surface is analyzed, showing that nonradiative recombination occurs near the sidewalls. This information suggests that the fabrication process could have detected and repaired these problems earlier.