1 Introduction

Self-emissive Micro-LED display has become one of the hotspots in research and industry owing to its many advantages, such as high brightness, high resolution, high contrast, fast response, long lifetime and so on [1,2,3,4,5,6,7,8,9,10,11,12,13]. Being a revolutionary display technology with high potential for mass production, it is considered to be one of the most promising future display technologies [3, 4, 8]. There are generally two approaches to achieve full color RGB Micro-LED display: RGB pixels consist of red/green/blue (RGB) Micro-LED chips [14, 15] and RG pixels from color conversion [5, 10, 11]. In the first approach, each pixel contains a set of red, green and blue Micro-LED chips. All these RGB chips are separately assembled on the same substrate through mass transfer. It has some major challenging problems to be solved before entering mass production. The accuracy and efficiency of the mass transfer are yet to be improved [16,17,18]. And it has to reduce the high power consumption. This is almost an insurmountable challenge for full-color RGB Micro-LED display due to the relative low EQE (external quantum efficiency) of the red Micro-LED chips [19, 20]. In comparison, color conversion Micro-LED display is much closer to mass production. It has blue/UV LEDs to pump the color conversion layer. It has many advantages such as fewer MURA defect, simpler chip mass transfer process and minimal electrical driving system. Thus color conversion has become a popular method to achieve full color Micro-LED display [21, 22]. On the other hand, QDs have extraordinary properties such as size-dependent energy gap, narrow emission spectrum, high luminescence efficiency, large stokes shift and less reabsorption [23,24,25,26,27]. It has become one of the crucial building blocks of color conversion [11, 22, 28, 29]. The millions of RGB sub-pixels on a display panel can be realized via incorporating QDCC film into Micro-LED back panel. This method not only improves the color rendering, brightness and gamut, but also helps to reduce the power consumption. Besides, the driving circuit is simpler and the cycle of massive transfer is also shortened [30].

However, the side light emission of Micro LED chips is strong and it is a problem to be solved before any display application, because it causes serious crosstalk in QDCC-LED devices. Ideally, all the light from LEDs should be directed to the front side of the panel. However, it is very difficult to achieve this through just setting reflective layer on the LED chips or the LED backplanes. Another problem to be solved is the overlap between the micro LED emission spectrum and the green color filter (G-CF) transmission spectrum. This causes excess blue light pass through G-CF. Both problems seriously reduce the color gamut [31]. In this work, we propose a simple and effective way to reduce crosstalk caused by LED side light emission. A double-layer bank was fabricated on the side of QDCC. The first layer of the bank constrained the quantum dots to obtain a flat morphology. The second layer of the bank was used to embed micro LEDs so that the side light emission of the LED was directed into the front, thereby greatly reduced the crosstalk. Moreover, we chose blue micro LED backlight with shorter wave length to minimize the overlap between the micro LED emission spectrum and the G-CF transmission spectrum. With these approaches, the color gamut of our QDCC-LED display reaches 107.53% NTSC.

2 Experimental Details

2.1 Preparation of Color Filter and Gray Banks

First, glass substrates were ultrasonically cleaned followed by ultraviolet ozone treatment. Next, patterned red, green and blue color resists were fabricated by conventional photolithography procedures. Black matrix resist was then coated onto the glass and baked at 230 °C. A buffer layer composed of a mixture of SiNx and SiOx was deposited on top of the resist by chemical vapor deposition (CVD) to provide an appropriate surface for subsequent film formation. Finally, the double layer banks were fabricated by conventional photolithography procedures.

2.2 IJP of QDs Inks

In this work, all QDs (CdSe/ZnS) were purchased from Suzhou Xingshuo Nanotech Co., Ltd. The weight concentrations of these two types QD inks were both ~ 25% with the solvent of cyclohexylbenzene. As for the scatter ink for blue pixels, the QD component was replaced with TiO2 spheres (diameter≈ 100 nm). The printing of QD inks was performed with Omnijet 200 (Unijet) printer. Printed QDs inks were then solidified under UV lamp (λ =  ~ 365 nm) for 5 min. All of these processes were performed under Nitrogen environment.

2.3 Characterization

Optical and fluorescent images of samples were obtained with a laser confocal microscope (Olympus FV1000-IX81, Japan). Fluorescence spectra were measured by a spectrometer (Princeton Instruments Acton SP2300, USA) equipped with a CCD camera (Princeton Instruments Pixis100, USA). Scanning Electron Microscope (SEM) images were obtained using Nova Nano SEM 200 (FEI).

3 Results and Discussion

Currently, two main techniques have been used to process and integrate the QDCC and Micro-LED backplane: one is overlaying the QDCC layer on the Micro-LED backplane directly, which has better efficiency and less inter-pixel crosstalk but needs fussy and complicated processing; the other is bonding a QDCC substrate onto the Micro-LED backplane, which is facile and low-cost. Considering its process and cost, we chose the latter technique in this work.

Gray bank material was chosen to enhance blue light reflection while having enough optical density (OD) to reduce light penetration. In this bank material, a large number of TiO2 nanoparticles (NPs) with size of ~ 200 nm were added to scatter light, as shown in Fig. S1b. The thickness of the bank was set to be ≥ 20 μm to reserve enough space for quantum dots and micro LED. The height of the micro LED and eutectic layer together was about 8 μm. The thickness of QDCC layer needs to be larger than 8 μm to obtain good color conversion efficiency and minimize the blue light leakage according to the QDs supplier data [32]. The bank was built by two step photolithography. Each lithography process formed a film of ~ 10 μm.

The QDCC films were fabricated by inkjet printing on the QDCC substrate. Inkjet printing is a new technology with the advantages of low cost, high degree of automation and small work load [33,34,35,36]. The red and green inks were prepared with core/shell CdSe/ZnS QDs. The blue ink was formed by dispersing scattering particles into acrylic resin.

Two schemes for ink-jet printing QDCC layer were designed. Figure 1 schematically depicts the fabrication process flow. The difference between the two schemes lies in the order of printing QDs. For scheme (a), the QD ink-jet printing process was done after the double-layer bank fabrication. For scheme (b), the QD ink-jet printing was done between the first bank layer and the second bank layer. In scheme (b), a 500 nm SiNx encapsulation layer was added to protect the QDs from the photolithography process of the second bank layer and the water and oxygen from atmosphere. The second bank layer was fabricated with lower baking temperature to prevent the QDs from heating failure.

Fig. 1
figure 1

The process diagrams of the fabrication of colorful QDCC by IJP process for a scheme (a) and b scheme (b)

Full size image

The sub-pixels of the 228 PPI 1.11 inch panel were arranged in a zigzag arrangement. The distance between each sub-pixel was 20 μm. G and B sub-pixels had the same opening area of 42*34 μm. The opening area of R sub-pixel was 52*27 μm. This pixel arrangement is shown in Fig. 2g.

Fig. 2
figure 2

The SEM images of QDCC structure: a red and d green QDs films before optimization for scheme (a); b red and e green QDs films after optimization for scheme (a); c red and f green QDs films for scheme (b). g The SEM images of QDCC R/G/B pixels

Full size image

For scheme (a), the printed QDs showed very uneven morphology. This is due to the high wetting ability of the QDs on the gray bank. In order to preserve enough space for micro LEDs, we set the pinning line of the printed R-QDs and G-QDs at a distance of 8 μm from the upper surface, as shown in Fig. 2a and d. However, the central film thickness we got in scheme (a) is only 3.76 and 4.07 μm respectively, which was too thin to convert all the blue light from micro LEDs to red and green light. This would lead to the leakage of blue light through QDCC films [37]. To optimize, we added more ink so that the central films thickness of R-QDs and G-QDs was increased to 9.59 and 8.55 μm respectively, as shown in Fig. 2b and e. But the edge of QD film climbed onto the top surface of the bank and encroached into the reserved space for micro LEDs. This resulted in bigger cell gaps and caused crosstalk. For scheme (b), the QD film was ink-jet printed after the first layer of the double-bank was fabricated. The QDs fully filled the pixel pit. The central thicknesses of the R-QDs and G-QDs were 14.4 and 15.46 μm respectively, as shown in Fig. 2c and f. The surfaces of these two QD films were very smooth. After the fabrication of the second bank layer, there were spaces of 11.95 and 10.72 μm in height reserved for the Micro LEDs.

We first used blue Micro LEDs with wavelength of 467 nm to pump the R-QDs and G-QDs. Table 1 summarizes the device performance with scheme (a) and (b). The cell gaps of the integrated device were 2.481 μm and 0.377 um for scheme (a) and (b) respectively. The cell gap of scheme (a) device was much higher than scheme (b) device because the edge of QD film climbed too high in scheme (a), which hindered the insertion of blue Micro LEDs. It is worth to point out that in the QDCC device, as Fig. S1a shows, there are four main sources that can cause light crosstalk: 1. Blue light travels across cell gaps between the array substrate and bank to hit the adjacent bank and then be scattered; 2. A small part of blue light can go directly into the adjacent sub pixels through the bank; 3. Blue light gets reflected by the metal materials in the array substrate; 4. There is some interface reflection when some blue light hits the lower surface of array substrate. To reduce the second type of crosstalk, we set the distance between sub-pixels to be 20 μm, which effectively blocked all light getting through the bank wall. As shown in Fig. S1b, the transmittance of 10 μm thick bank for blue, green and red light were 4, 4.44 and 5.76% respectively measured by the full bank film without graphics fabricated on the bare glass. From the tested data of 10 μm thick bank, we could simply calculate that the transmittance of 20 μm thick bank for blue, green and red light were 4%*4% = 0.16%, 4.44%*4.44% = 0.20% and 5.76% *5.76% = 0.33% respectively. As for the third and fourth type of crosstalk, the reflected light was relatively weak. Moreover, all the devices had the same Micro LED TFT arrays. Thus we believe these type of crosstalk had little and similar influence on device performance. The main part of crosstalk came from the first source. As mentioned above, the side light emission of Micro LED chips was very strong. The larger the cell gap was, the more blue back light traveled across the gap, hit the adjacent bank and scattered into adjacent QDs then lighted the adjacent sub-pixels. This phenomenon resulted in unwanted red or green light emitted when only blue light was needed, for instance. Consequently, large cell gaps caused severe crosstalk in scheme (a) compare to scheme (b). A comparison of red, green, blue luminance spectra between the schemes is shown in Fig. 3a–c. Severe crosstalk was found in red, green and blue screens of scheme (a) rather than in those of scheme (b). Take the red screen for example, the peaks of blue and green light of 467 nm and 532 nm were very high for the device of scheme (a), while little blue and green light was found in the spectrum for the device of scheme (b). This phenomenon was also found in green and blue screens. The same difference was also observed in microscope pictures, as shown in Fig. 3d–i. Here, unwanted colors can be found in the luminance spectra and microscope pictures of scheme (a). Table S1 summarizes and compares the light leakage ratio of the two type devices. The light leakage ratio was defined as the proportion of unwanted light color brightness in the total luminous brightness (R: 580–780 nm; G: 500–580 nm; B: 380–500 nm). For the device of scheme (b), the light leakage ratios of red, green and blue screens were 1.34%, 7.87% and 7.34% respectively, which were much lower than that of scheme (a) with 18.96%, 16.05% and 11.44% light leakage ratios. The lower light leakage ration resulted in much higher color gamut for scheme (b) device. The NTSC of scheme (b) device is 91.59% while the NTSC of scheme (a) device is only 75.44%. In the microscope picture of illuminated scheme (a) devices, uneven light emission within the sub-pixel can be seen in the red and green arrays. Since the thicknesses of the central R-QDs and G-QDs films were thinner than the edge parts, the middle part of the sub-pixel was less bright.

Table 1 The performance summaries of scheme (a) and scheme (b) devices
Full size table
Fig. 3
figure 3

The comparison of lighted spectra between scheme (a) and scheme (b) devices: a red color; b green color; c blue color. The comparison of lighted microscope pictures between scheme (a) and scheme (b) devices: d red color, e green color and f blue color for scheme (a) device; g red color, h green color and i blue color for scheme (b) device

Full size image

Although better than scheme (a), the CIE-y of the green color device for scheme (b) was still relatively low and cannot meet the wide color gamut requirement. As shown in Fig. 3b, a blue color peak about 480 nm was found in the green luminance spectrum. This blue peak mainly comes from the transmitted light from the blue LEDs (BLEDs). Here we need to discuss about the function of color filter (CF). Generally, CF does two things: the first is helping to reduce the excitation of QDs films by ambient light so that it improves the contrast of the display. The second is taking out the unabsorbed blue lights passing through QD films to further improve the color gamut of the display [38]. Actually, the thickness of our QD films was about 10 μm. It absorbed about 90% of the blue light. But the transmission spectrum of our G-CF was relatively wide and partially overlapped with a part of the emitting spectrum of the BLEDs. Some of the BLED backlights directly passed through the G-CF. Consequently, emission wavelength of the backlight played a crucial role in optimizing the performance. We needed to eliminate the overlap between the emitting spectrum of the BLEDs and the transmission spectrum of the G-CF. Figure 4a and b show the transmittance spectrum of the G-CF, the transmittance spectrum of the R-CF and the emitting spectrum of two different BLEDs with central wavelength at 467 and 455 nm respectively. The emitting spectra of these two BLED is also shown in Fig. S2. The penetration rates of backlights in G-CF were 17.10% for the 467 nm BLED and 8.20% for the 455 nm BLED. The higher the penetration rate, the more blue light go through the G-CF in the green sub-pixel. Thus shifting the BLED emission peak from 467 to 455 nm greatly reduced the blue light leakage from G-CF.

Fig. 4
figure 4

The long-wavelength a and short-wavelength b BLED spectra and the transmission spectra of R/G-CF

Full size image

Shifting the BLED emission peak from 467 to 455 nm also improved the brightness of G-QDCC. In Fig. 5a and b, the spectra of QDCC devices excited by blue backlights with different wavelengths were compared. The excited fluorescence spectra of R-QD films were same, because the red QDs completely filtered out the blue lights. In contrast, there were different levels of a blue peak in the two fluorescence spectra of G-QD films with different blue backlights wavelengths. The intensity of this remnant BLED backlight peak with wavelength of 467 nm was much higher than that with wavelength of 455 nm. The transmissivity of blue backlight in green QDs film were 7.15% and 3.63% for long-wavelength and short-wavelength respectively, as shown in Table S2. Microscopic images of red and green pixels are exhibited in the insets. The center of the pixels in the green image is slightly cyan for the 467 nm sample. This agrees with the higher blue light penetration in this case. As a comparison, pixels in the 455 nm sample showed much better quality. Adopting the 455 nm backlight helped to increase the CIE-y of the green screens from 0.5871 to 0.6635, as shown in Table S3. Figure 5c shows the CIE chromaticity coordinates of QDCC with blue LED backlights of different wavelengths. Benefited from the low transmissivity and high luminescence conversion efficiency with short wavelength BLEDs, the maximum color gamut was increased to 107.53% of NTSC from the 91.59%. Pictures of QDCC-LEDs with different blue backlight wavelengths are exhibited in Fig. 5d. It can be seen that the picture with short-wavelength blue backlight are brighter and more delicate than that with long-wavelength blue backlight.

Fig. 5
figure 5

a and b are the spectra of red and green QD films excited by blue backlight at different wavelengths, the insets in upper right corner of figure show red and green microscopic images. c The color gamut of QDCC-LED display with long blue backlight wavelength (yellow triangle area), and short blue backlight wavelengths (black triangle area). d The physical photo of full-color QDCC devices display (size: 1.1 inch) with different blue backlight wavelengths

Full size image

Light conversion efficiency (LCE) is calculated as Lout/Lin, where Lout is the luminance intensity of the converted light, e.g., green and red light from QD. Lin is the luminance intensity of the input light from BLEDs. LCE characterizes how much blue emission is converted by the QDs [39,40,41]. The LCEs of our R-QDs and G-QDs were 1.58 and 2.1 respectively for the 455 nm blue backlight. The LCEs of R-QDs and G-QDs with 467 nm blue backlight were 0.78 and 2 respectively. This clearly shows that with the same backlight brightness, the short-wavelength blue backlight excited QDs to emit more light.

Luminance intensity and color deviation vary with temperature. Such temperature dependence of these properties in QDCC-LED red, green and blue devices and RGB Micro-LEDs are shown in Fig. 6. As the temperature raise from 0 to 80 °C, the luminance of all devices decreased. The temperature induced luminance attenuation was most serious in the Micro LEDs in red display. It was more serious than QDCC devices. This is because of the intrinsic unstable character of red LED chips. Big efficiency roll off inevitably happened when the temperature rises. For green and blue, QDCC devices with 455 and 467 nm BLED backlight exhibit attenuations closer to those in Micro LEDs. Among them, the green QDCC device with 455 nm BLED backlight showed bigger attenuation than the 467 nm device and the green Micro LED. This is due to an encapsulation failure. In some of the G-QD sub-pixels, about 300 nm gully was found between the G-QDs and the gray bank. This led to more serious efficiency roll off under heating, as shown in Fig. S3a.

Fig. 6
figure 6

Luminance a-c and color deviation d–f variation with temperature from (0 to 80 °C) of QDCC devices with different wavelength blue Micro LED (455, 467 nm) and Micro-LED in red, green and blue display

Full size image

Color deviation (∆E) under different temperature was calculated as:

$$\Delta E=\sqrt{{\left(x-{x}_{0}\right)}^{2}+{\left(y-{y}_{0}\right)}^{2}}$$

where \({x}_{0}\), \({y}_{0}\) represent the color coordinates at 0 °C. \(x\), \(y\) are the color coordinates at various temperature. For red and green displays, the color deviations of QDCC devices with 455 nm BLED backlight are small and close to that in the Micro LED devices. For blue displays, color deviation in QDCC device was even better than that of the blue Micro LED device. In this case, the QDCC device with 467 nm BLED backlight also showed smaller color deviation than the Micro LED device thanking to the color filter which narrowed the emission spectra. Yet the QDCC device with 455 nm BLED backlight still was better. This is also an effect of the color conversion and filtering. As shown in Fig. S3b, when the temperature raise from 0 to 80 °C, the peak positions of the blue Micro LEDs red shifted by 3 nm for both BLEDs. However, as mentioned above, there is an overlap between BLED emission spectrum and G-CF transmission spectrum, especially for the 467 nm case. The red shift in the BLED emission causes more blue light transmission. Thus the device with 455 nm BLED backlight showed smaller color change. To conclude, QDCC devices with short wavelength blue backlight had better temperature stability and they were also superior to corresponding Micro-LEDs.

4 Conclusions

In summary, we realized an active matrix full-color display by combining QD enhanced CCL with Micro-LEDs, where Micro-LEDs acted as blue sub-pixels as well as the excitation light for the red and green conversion QDs. Patterned QD layers were prepared by IJP, which is a low-cost and effective approach for patterned QD films. By using double-layer bank and replacing the blue backlight of 467 nm with 455 nm, the performance of QDCC-LEDs were effectively improved. A QDCC-LED display panel with resolution of 228 PPI and size of 1.11 inch was fabricated. It has a color gamut of 107.53% NTSC.