Open Access
Add to BibSonomy
Abstract
Ultraviolet micro-LEDs show great potential as a light source for maskless photolithography. However, there are few reports on micro-LED based maskless photolithography systems, and the studies on the effects of system parameters on exposure patterns are still lacking. Hence, we developed a maskless photolithography system that employs micro-LEDs with peak wavelength 375 nm to produce micrometer-sized exposure patterns in photoresists. We also systematically explored the effects of exposure time and current density of micro-LED on static direct writing patterns, as well as the effects of stage velocity and current pulse width on dynamic direct writing patterns. Furthermore, reducing the size of micro-LED pixels enables obtaining high-resolution exposure patterns, but this approach will bring technical challenges and high costs. Therefore, this paper proposes an oblique direct writing method that, instead of reducing the micro-LED pixel size, improves the pattern resolution by changing the tilt angle of the sample. The experimental results show that the linewidths of the exposed lines decreased by 4.0% and 15.2%, respectively, as the sample tilt angle increased from 0° to 15° and 30°, which confirms the feasibility of the proposed method to improve the pattern resolution. This method is also expected to correct the exposure pattern error caused by optical distortion of the lens in the photolithography system. The system and method reported can be applied in various fields such as PCBs, photovoltaics, solar cells, and MEMS.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Photolithography is a key fundamental process in micro/nano fabrication and has extensive applications in several industries, including semiconductors, micro-electromechanical systems (MEMS), solar cells, printed circuit boards (PCBs), photovoltaics, and integrated circuits (ICs) [1–4]. Traditional photolithography techniques rely on photomasks, which are typically blank pieces of glass or quartz with a patterned metallic thin film. However, manufacturing these photomasks is not only costly, but also time-consuming and inflexible, hindering the progress of traditional photolithography techniques [5]. As a result, maskless photolithography approaches have gained great popularity due to their low cost and high flexibility. In recent years, various maskless photolithography techniques have been developed, such as focusing ion beam lithography [6], single photon lithography [7], electron-beam lithography [8], inkjet printing [9], laser direct writing [10], ultraviolet light-emitting diodes (UV-LEDs) direct writing [11], and digital micromirror devices (DMDs) and other spatial light modulators (SLMs) based methods [12]. Some of these technologies have already been applied to industrial manufacturing processes.
Compared to traditional light sources, LED is a new type of semiconductor light source that has attracted considerable attention in both industry and academia for its exceptional performance, including high efficiency, low power consumption, long lifetime, and environmental friendliness. Additionally, LED is capable of emitting UV light with wavelengths of approximately 365 nm (i-line), 405 nm (h-line), or even shorter, which is comparable to the light emitted by mercury lamps. Therefore, as an alternative to mercury lamps, UV-LEDs are well-suited for use in photolithography systems. Guijt et al have developed a direct writing lithography system using UV-LEDs, providing a high-resolution, cost-effective solution for maskless photolithography [11]. Besides the advantages of UV-LEDs mentioned above, UV micro-LEDs can provide arrays of light emitting pixels on small-scale chips. UV micro-LEDs are expected to achieve rapid exposure of large-area micro/nano patterns by controlling their pixels to produce different emission patterns to meet the demand for high productivity in PCBs and other semiconductor manufacturing. This makes them a great potential light source for maskless photolithography compared to the low production efficiency of laser direct writing machines.
At present, there are only a few reports on direct writing photolithography system using UV micro-LEDs. For static direct writing, Elfstrom et al developed a micro-photolithography system utilizing CMOS-driven micro-LED arrays to produce exposure patterns consisting of exposed dots with diameters of 8–9 µm [13]. Stonehouse et al have constructed a maskless photolithography system based on a micro-LED array with peak wavelength 405 nm that could produce exposed dots of approximately 40–45 µm in the photoresist material [14]. For dynamic direct writing, Elfstrom et al fabricated 8 µm exposed lines using photolithography system based on a 370 nm micro-LED array at a pulse width of 40 ns and a velocity of 5 µm/s [13]. Guilhabert et al achieved features as small as 0.5 µm using a 14 µm-diameter UV micro-LED pixel at a speed of 140 µm/s [15]. System parameters are important indicators for measuring the performance of lithography systems. However, systematic studies of the effects of system parameters on the static and dynamic direct writing patterns of maskless photolithography systems based on micro-LED arrays have not been reported.
In this work, we prepared a 375 nm-emitting micro-LED array device and constructed a maskless photolithography system using this device, capable of creating micrometer-sized exposure patterns in photoresist. Firstly, the intensity distribution of projected spots in the system was analyzed. We also systematically investigated the effects of exposure time and current density of micro-LED on the exposure patterns in static direct writing, as well as the effects of stage velocity and current pulse width on the exposure patterns in dynamic direct writing. The size of the exposure patterns increases with the increase of exposure time, current density and current pulse width, and decreases with the increase of stage velocity. The research results are significant for analyzing and optimizing the performance of micro-LED based maskless photolithography systems. In addition, as an important index of the photolithography system, the higher the resolution of the exposure pattern, the smaller the chip components that can be manufactured. We can achieve smaller feature sizes by reducing the size of the micro-LED pixels to improve the pattern resolution [15]. However, fabricating micro-LED arrays with smaller pixel sizes is challenging and expensive. Hence, the paper proposes a novel oblique writing method as an alternative to reduce the size of micro-LED pixels, which enhances pattern resolution by varying the tilt angle of the sample. The experimental results indicate that the proposed method effectively reduced the size of the exposed lines, thereby improving pattern resolution.
2. System structure and characterization
2.1 UV micro-LED device
Figure 1(a) shows a typical schematic diagram of a GaN-based micro-LED, where the substrate for the UV micro-LEDs is sapphire. Micro-LEDs were prepared on 375 nm-emitting epitaxial wafers grown on a patterned sapphire substrate. Each micro-LED is composed of several layers including AlGaN/GaN buffer layer, n-GaN layer, p-GaN layer, multiple quantum wells (MQWs) layer, electron blocking layer (EBL), indium tin oxide (ITO) thin film layer and SiO2 passivation layer. The n-pad and p-pad consist of a Ti/Au bilayer and are utilized as metal electrodes. The micro-LED chip was bonded to a PCB through wire bonding. The process of micro-LED device fabrication is detailed in our previous research results [16,17]. The diameter of the p-pad layer of 40 µm was calibrated to be the pixel size as it was designed to be the same size as the mesa.
Fig. 1. (a) Schematic diagram for chip structure of a 40 µm micro-LED. (b) Normalized EL spectra of micro-LED at different current densities of 80, 400, 800, and 1600 A/cm2, respectively. (c) I-V and P-I characteristic curves of micro-LED.
A source measurement unit (SMU, Keithley 2614B) was used to drive the micro-LED device by direct current (DC) and pulse modes. The normalized EL spectra of micro-LEDs were measured by an integrating sphere (IS/LED 25). From Fig. 1(b), the micro-LED device with 40 µm pixels emitted at a peak wavelength of about 375 nm at various current densities from 80A/cm2 to 1600 A/cm2. When the current density was below 1600A/cm2, the peak wavelength of the electroluminescence (EL) spectra changed very little, while the full width at half maximum (FWHM) increased slightly with increasing current density, which indicated that the micro-LED had good monochromaticity [18]. As shown in Fig. 1(c), the current versus voltage (I-V) curve of the micro-LED device was obtained using the high-precision SMU (Keithley 2614B). It was found that the 40 µm micro-LED exhibited a leakage current in the order of nA, proving its high quality [19]. The light output power versus current (P-I) curve was measured through the integrating sphere (IS/LED 25). Figure 1(c) also illustrated that light output power of micro-LED approximately increased linearly with current. In DC mode, we measured the optical power of the micro-LED at different currents of 1, 5, 10, and 20 mA to be 156.9, 691.6, 1240.9, and 2246.7 µW, respectively.
2.2 System structure
Figure 2 shows the schematic diagram and image of the photolithography system. The system includes components such as micro-LED device, yellow light (Hayear HY-D3520), coated plano convex lens (Thorlabs LA1131-A), aspheric condenser lens (Thorlabs ACL25416U-A), 50-50 beam splitters (Thorlabs EBS1), tube lens (Thorlabs TTL180-A), CCD camera (Daheng MER-504-10GM/C-P), computer, microscope objective, and XYZ stage. The XYZ stage can be a manual XYZ stage, or be a combination of a motorized XY stage (Thorlabs M30XY/M) and a manual translation stage.
Fig. 2. (a) Schematic diagram, and (b) image of the setup during direct writing showing the micro-LED, plano convex lens, beam splitter, aspheric condenser lens, tube lens, microscope objective, yellow light, CCD camera, XYZ stage, and the sample.
As shown in Fig. 2, the micro-LED device was installed on a manual XYZ stage. The UV light emitted from the micro-LED device was collected and collimated by a coated plano convex lens (Thorlabs LA1131-A). Subsequently, the light was added to the main optical path by a 50:50 beam splitter (Thorlabs EBS1) and focused by either a 4X-40X infinity-corrected microscope objective (Phenix 195 Achromatic Objective Series) or a 50X long working distance microscope objective (Rixin LPL 50X). In this way, the projected spots were obtained on the surface of the sample mounted on the XYZ stage. The computer-controlled motorized XY stage (Thorlabs M30XY/M) allowed for precise movement of the sample mounted on the stage, which could move with a minimum motion increment of 2.5 µm and a maximum rate of 2.4 mm/s with a repeatability of approximately ±1 µm. The manual translation stage enabled a minimum displacement of 10 µm with an accuracy of 5 µm. Therefore, the combined XYZ stage could precisely control the focus on the sample. A CCD camera (Daheng MER-504-10GM/C-P) with an accuracy of 3.45 µm was used to observe the sample or the projected spot. A tube lens (Thorlabs TTL180-A) was used to control the field of view (FOV) of the CCD camera. The yellow light (Hayear HY-D3520) could illuminate the system. The light emitted from it could be collected by an aspheric condenser lens (Thorlabs ACL25416U-A) and added to the main optical path through a 50:50 beam splitter.
In addition, the computer could simultaneously control both the micro-LED device and the motorized XY stage, enabling automatic regulation of the exposure dose and exposure area of the sample. The system also permitted real-time data analysis and manipulation.
2.3 System characterization
The light emitted by the micro-LED could be focused through a microscope objective to form a spot, which was then projected onto the sample. The optical power meter (Thorlabs PM100) was placed at the sample to measure the optical power of the projected spot. The optical path should be designed to minimize optical power loss, thereby reducing the overall power consumption of the system. For the 40 µm micro-LED, the optical power of its projected spot was measured to be 1.3 µW using a 4X microscope objective with an NA of 0.1, 0.9 µW using a 10X microscope objective with an NA of 0.25, and 0.4 µW using a 40X microscope objective with an NA of 0.65 at the current density of 80 A/cm2. As shown in Fig. 3, the optical power of the projected spot decreased almost linearly as the magnification of the microscope objective increased. Related research shows that the beam splitter in the whole photolithography system can cause significant optical power loss, while the transmission loss of the objective lens is relatively small [13]. In this work, we found that the magnification of the microscope objective also affected the projected optical power. With the increase of the magnification of the microscope objective, it will cause a certain amount of optical power loss.
Fig. 3. Optical power of the projected spot at different microscope objective magnifications of 4X, 10X, and 40X, respectively.
What is more, although the 40X microscope objective has a lower projected optical power, but it also has a larger NA value. The resolution of a photolithography system can be expressed using the Rayleigh’s equation [20]:
where R is the feature size, λ is the wavelength of the light source, NA is the numerical aperture, and k1 is the unit less Rayleigh constant. According to Eq. (1), the larger the NA value, the smaller the feature size and the higher the resolution. Therefore, to get higher resolution, it is more appropriate to select a 40X microscope objective for the system.Then, the optical powers of the projected spot of the 40 µm micro-LED through the 40X microscope objective were measured at current densities of 160, 240, 320, and 400 A/cm2, respectively, as shown in Table 1. The measured values were 0.5, 0.7, 0.9, and 1.1 µW, respectively. Furthermore, the CCD camera (Daheng MER-504-10GM/C-P) was used to image the projected spot at different current densities. The CCD image of the projected spot at a current density of 160 A/cm2 was first analyzed using a MATLAB program to derive a 3D map of the relative intensity distribution of the projected spot in Fig. 4(a). We also extracted a 2D map of the relative intensity distribution in the FWHM region, as shown in Fig. 4(b). The relative intensity ratio of the FWHM area to the whole projected spot could be calculated to be 11.3% by integration using the MATLAB program, and thus the optical power in the FWHM area was calculated to be about 0.06 µW. Using this method, we also calculated that the optical powers in the FWHM area of the projected spot at the current densities of 240, 320, and 400 A/cm2 were 0.08, 0.11, and 0.13µW, respectively, as shown in Table 1.
Fig. 4. (a) The 3D map of the relative intensity distribution of the projected spot, and (b) the 2D map of the relative intensity distribution of the FWHM area at a current density of 160 A/cm2.
Table 1. Optoelectronic parameters for a 40 µm micro-LED
Moreover, Figs. 5(a)-(d) show the CCD images of the projected spot at current densities from 160 A/cm2 to 400 A/cm2, and the intensity profiles along the x-axis and y-axis were measured on these images using the MATLAB program, as shown in Figs. 5(e)-(h) and Figs. 5(i)-(l), respectively. It is found that the relative intensity is high in the central area of the projected spot, and is low in the edge area. This intensity profile is similar to a Gaussian distribution, which can help to adjust the size of exposure patterns in photolithography. The spot size is typically determined at 13.5% of the peak intensity of the intensity curve, which is also referred to as the 1/e2 diameter. The area within the 1/e2 diameter contains the majority of the spot power [21–23]. The 1/e2 beam diameters of the projected spots on the x- and y-axis for current densities of 160, 240, 320, and 400 A/cm2 were derived from intensity curves shown in Figs. 5(e)-(l). The 1/e2 diameters of the x-axis were 23.3, 24.1, 25.9, and 26.7 µm, respectively, and those of the y-axis were 23.4, 23.6, 24.0, and 25.7 µm, respectively. It was assumed that the relative intensity distribution of projected spot was uniform. We could calculate the average 1/e2 diameters and power densities of the projected spots as listed in Table 1. Furthermore, the FWHM values for the x- and y- axis of the projected spot were derived at current densities of 160, 240, 320, and 400 A/cm2. The FWHM values for the x-axis were 6.3, 6.6, 6.7, and 6.9 µm, respectively, and those for the y-axis were 6.2, 6.4, 6.6, and 6.9 µm, respectively. As a result, Table 1 shows the calculated average FWHM values and power densities of FWHM areas. At the same current density, the 1/e2 diameters of the x-axis and y-axis were quite close to each other, and their FWHM values were also similar, indicating that the shape and intensity of the projected spot were relatively uniform.
Fig. 5. (a)-(d) CCD images of the projected spot at current densities from 160 A/cm2 to 400 A/cm2, (e)-(h) x-axis relative intensity curves of the projected spot with 1/e2 diameters of 23.3, 24.1, 25.9, and 26.7 µm, and the FWHM values of 6.3, 6.6, 6.7, and 6.9 µm, and (i)-(l) y-axis relative intensity curves of the projected spot with 1/e2 diameters of 23.4, 23.6, 24.0, and 25.7 µm, and the FWHM values of 6.2, 6.4, 6.6, and 6.9 µm, at current densities of 160, 240, 320, and 400 A/cm2, respectively.
2.4 Photolithography experiments
To test the exposure capability of the photolithography system, we exposed photoresist (Shipley Microposit S1818) using a 40 µm micro-LED device. Microposit S1818 photoresist can be exposed using a light source with a wavelength range of 350–450 nm. Therefore, it is suitable for exposure at a wavelength of 375 nm, although its exposure characteristics are best at 436 nm. The 4-inch silicon wafer was diced into individual 1.5 × 1.5 cm2 silicon substrates. The silicon substrates were then cleaned with ethanol in an ultrasonic bath and rinsed with deionized water. The surface water was removed with N2 gas, and the silicon substrates were then dried on a hot plate at 115 °C for 5 s. The silicon substrate was coated with an approximately 2 µm thick Microposit S1818 film by coating it at 600 rpm for 6 s and subsequently at 4000 rpm for 30 s. The film-coated substrate was then soft-baked by heating it on a hotplate at 115 °C for 1.5 min prior to exposure. The substrate was transferred to the XYZ stage of the system and exposed using the micro-LED device. After exposure, a 2.38% concentration of tetramethylammonium hydroxide (TMAH) solution was used for development. The substrate was immersed in the developer solution and slowly moved for 60–90 s. The residual developer was then rinsed from the substrate with deionized water. Finally, the surface water of the substrate was dried with N2 gas.
3. Results and discussion
3.1 Static direct writing
The exposure dose is a critical control factor in the photolithography process, directly affecting the size and quality of the exposure patterns [24]. In static direct writing, the exposure dose is determined by the optical power per unit area (i.e. power density) and the exposure time [13]. First, the effect of exposure time on the pattern exposed by the micro-LED based maskless photolithography system was investigated.
In this experiment, we used a 40 µm micro-LED to expose S1818 photoresist through a 40X microscope objective. Based on the data presented in Table 1, at a current density of 160 A/cm2, the power densities of the projected spot and the FWHM area were 116.3 and 192.6 mW/cm2, respectively. Therefore, the exposure doses of the projected spot were approximately 116.3, 232.6, and 581.5 mJ/cm2 for 1, 2, and 5 s, respectively. The exposure doses of the FWHM area were about 192.6, 385.2, and 963.0 mJ/cm2 for 1, 2, and 5 s, respectively. The micrographs in Figs. 6(a)-(c) imaged by an optical microscope (Nikon ECLIPSE Ci-S) show exposed dots exposed for different exposure times. Since the incompletely developed black borders still contained photoresist that could be removed in post-processing, we measured the size of the fully developed area in the exposed patterns as the pattern size. As shown in Figs. 6(a)-(c), the diameters of exposed dots exposed for 1, 2, and 5s were measured to be about 6.2, 8.8, and 20.3 µm, respectively. It is observed that the diameter of the exposed dot increases as the exposure time increases. The reason is that the exposure dose and the number of UV photons causing a molecular reaction in the photoresist are proportional to the exposure time [7]. At an exposure time of 1 s, only the exposure dose in the FWHM area was high enough to generate sufficient numbers of UV photons to fully react with the photoresist molecules, thus the diameter of the exposed dot was about 6.2 µm, which was almost equal to the average FWHM value of 6.3 µm shown in Table 1. An increase in exposure time results in a higher exposure dose of the projected spot. Hence, a larger area of the projected spot had enough UV photons to react with the photoresist molecules, leading to an increase in the diameter of the exposed dot. When the exposure time reached 5 s, the diameter of the exposed dot increased to about 20.3 µm, which was close to the average 1/e2 diameter of the projected spot of 23.4 µm as listed in Table 1. The results indicate that the diameter of the exposed dot can be adjusted to vary from the FWHM value to the average 1/e2 diameter of the projected spot by controlling only the exposure time at the same current density. However, the edges of the exposed dots were not completely regular due to the uneven distribution of light intensity. What is more, we found that a black border appeared around the exposed dot. This was due to the high relative intensity in the central area of the projected spot and the low relative intensity in the edge area, as show in Fig. 5(e) and Fig. 5(i). The central area with a sufficient exposure dose to change the photoresist properties was fully developed. The area that had not fully developed formed a black boundary. The irregular edges and black borders of the exposure dots were both caused by poor focusing of the projected spot. Spot focusing can be improved by integrating micro-optics on the micro-LEDs [25]. As shown in Figs. 6(d)-(f), the heights of the exposed dots exposed for 1, 2, and 5 s were measured with a profilometry (Bruke Dektak XT) to be 2095, 2071, and 1911nm, respectively, which were the thickness of the photoresist film.
Fig. 6. Static direct writing: (a)-(c) micrographs of the exposed dots imaged by an optical microscope with diameters of 6.2, 8.8, and 20.3 µm, and (d)-(f) profile curves of exposed dots tested by a profilometer with heights of 2095, 2071, and 1911nm, exposed for 1, 2, and 5 s, respectively.
In addition to exposure time, current density also affects the output optical power of micro-LEDs and thus the exposure dose. In further experiments, we investigated the effect of micro-LED current density on the exposure pattern. A 40 µm micro-LED was also used to expose the S1818 photoresist at different current densities for 5 s through a 40X microscope objective. As shown in Table 1, the power densities of the projected spot were 116.3, 156.1, 183.4, and 204.1 mW/cm2 when the current densities of the micro-LED were 160, 240, 320, and 400 A/cm2, respectively. Therefore, the exposure doses of the projected spots for 5 s were 581.5, 780.5, 917.0, and 1020.5 mJ/cm2, respectively. Figures 7(a)-(d) illustrate micrographs of exposed dots obtained at various current densities. The exposed dot diameters were determined by the optical microscopy and were approximately 20.3, 22.5, 25.5, and 26.3 µm for current densities of 160, 240, 320, and 400 A/cm2, respectively, which closely matched the 1/e2 diameters of the projected spots. It has been observed that the diameter of the exposed dot increases as the micro-LED current density increases. As shown in Figs. 5(e)-(l), an increase in current density resulted in an increase in the relative intensity, FWHM value, and 1/e2 diameter of the projected spot. Consequently, for the same exposure time of 5 s, as the current density increased, the exposure dose also increased, and a larger area of the projected spot had a sufficient exposure dose to produce more UV photons to react with the photoresist molecules, leading to an increase in the diameter of the exposed dot. Additionally, the heights of the exposed dots were also measured using the profilometry, and the measured profile curves were shown in Fig. 7(e)-(h). The heights of the exposed dots were measured to be 1911, 1917, 1907, and 1905nm, which agreed with the thickness of the photoresist films, indicating that the exposure doses of 5 s at current densities ranging from 160 A/cm2 to 400 A/cm2 were sufficient to fully expose the photoresist.
Fig. 7. Static direct writing: (a)-(d) micrographs of exposed dots imaged by an optical microscope with diameters of 20.3, 22.5, 25.5, and 26.3 µm, and (e)-(h) profile curves of exposed dots tested by a profilometer with heights of 1911, 1917, 1907, and 1905nm, at current densities of 160, 240, 320, and 400 A/cm2, respectively.
However, increasing the diameter of the exposed dot can decrease the resolution of the exposure pattern. To improve pattern resolution, it is important to control the current density of the micro-LED in the photolithography system at the same exposure time. In static direct writing, a trade-off needs to be made between the current density of the micro-LED and the exposure time to optimize the exposure pattern for smaller feature size and higher resolution.
3.2 Dynamic direct writing
The experiments presented above tested the static direct writing capability of the system. Furthermore, by using a motorized XY stage (Thorlabs M30XY/M) with motion function enabled us to further explore the dynamic direct writing capability of the system. In the following experiment, a 40 µm micro-LED was used to expose the S1818 photoresist. The exposure was done through a 50X long working distance microscope objective at a current density of 800 A/cm2. Typically, dose energy is equal to optical power multiplied by exposure time in static direct writing. However, during dynamic direct writing, the XYZ platform on which the sample is mounted moves and the dose curve is calculated using the following equation [13]:
where P stands for the total optical power, R is the radius of the pixel, and v is the stage velocity. When a curve is exposed, such as a half-circular curve, y (≤R) is the coordinate perpendicular to the stage velocity. According to Eq. (2), the exposure dose varies inversely with the stage velocity. The samples were exposed by moving the motorized XY stage at velocities ranging from 10 µm/s to 100 µm/s. The measured value of P was 1.6 µW through the 50X microscope objective. Therefore, the maximum exposure doses were calculated using Eq. (2) to be 509.6, 101.9, and 51.0 mJ/cm2 when the samples were moved at the velocities of 10, 50, and 100 µm/s, respectively. Micrographs of lines exposed at different stage velocities are shown in Figs. 8(a)-(c). The optical microscopy was also used to measure the pattern linewidths. As shown in Figs. 8(a)-(c), the linewidths of the exposure lines were measured to be 15.1, 6.3, and 4.3 µm at velocities of 10, 50, and 100 µm/s, respectively. The exposed lines exhibited good uniformity, with heights of 2137, 2028, and 2020nm as measured by the profilometry, as shown in Figs. 8(d)-(f). It is observed that the linewidth of the exposed line decreases as the stage velocity increases. This is due to the inverse proportionality between the exposure dose and the stage velocity, as shown in Eq. (2). When the stage velocity is higher, the exposure dose that causes a photoresist reaction decreases, resulting in a reduction in the linewidth of the exposed line. Therefore, increasing the stage velocity facilitates reducing the size of exposed lines and improving their resolution.Fig. 8. Dynamic direct writing: (a)-(c) micrographs of exposed lines imaged by an optical microscope with linewidths of 15.1, 6.3, and 4.3 µm, and (d)-(f) profile curves of exposed lines tested by a profilometer with heights of 2137, 2028, and 2020nm, at stage velocities of 10, 50, and 100 µm/s, respectively.
In addition, the photolithography system employs a high-precision SMU (Keithley 2614B) capable of generating pulses. It can also be found from Eq. (2) that the total optical power of the projected spot is proportional to the exposure dose. In pulse mode, the output optical power of the micro-LED can be controlled by adjusting the current pulse width, which affects the total power of the projected spot and the exposure dose. This enables the investigation of the impact of varying current pulse widths on exposure patterns. During the experiment, the micro-LED emitted light to expose S1818 photoresist through a 50X objective lens at a current density of 800 A/cm2. The stage velocity was set to be 20 µm/s. The output optical power of the 40 µm micro-LED was controlled using pulses with a repetition rate of 50 Hz. The maximum exposure doses achieved were 63.7, 127.4, and 191.1 mJ/cm2 for current pulse widths of 5, 10, and 15 ms, respectively. Figures 9(a)-(c) show micrographs of lines exposed with different current pulse widths. The linewidths of the exposed lines were measured by optical microscopy and were 4.5, 6.9, and 10.2 µm for current pulse widths of 5, 10, and 15 ms, respectively. The heights of the lines were measured by profilometry and were found to be 2023, 2014, and 2006nm, as shown in Fig. 9(d)-(f). The results show that the linewidth of the exposed line increases as the current pulse width increases. This is because when the micro-LED device is driven with pulsed mode, the projected optical power increases linearly with the current pulse width, resulting in an increase in the exposure dose. In other words, we can obtain smaller-sized exposure patterns by reducing the current pulse width, which improves the resolution of the patterns. Moreover, when multiple micro-LED pixels are used for exposure, the variation in output optical power between different pixels can be adjusted with current pulse widths, thus helping to compensate for the efficiency of the photolithography system [13].
Fig. 9. Dynamic direct writing: (a)-(c) micrographs of exposed lines imaged by an optical microscope with linewidths of 4.5, 6.9, and 10.2 µm, and (d)-(f) profile curves of exposed lines tested by a profilometer with heights of 2023, 2014, and 2006nm, for current pulse widths of 5, 10, and 15 ms, respectively.
3.3 Oblique direct writing
The maskless photolithography system based on micro-LEDs can expose pixel array patterns directly on the photoresist or dynamically write lines on the photoresist by driving one or more micro-LED pixels. Therefore, we can obtain exposure patterns with smaller feature sizes and improve pattern resolution by reducing the pixel size of micro-LEDs. However, this approach increases the difficulty of fabricating micro-LED devices, which is not only technically challenging but can also be costly. In addition, the smaller size of micro-LEDs increases their surface recombination and sidewall damage, leading to reduced efficiency and thus increased power consumption [26]. Therefore, this paper proposes an oblique direct writing method to reduce the pattern size by changing the tilt angle of the sample, which does not require a reduction in the size of the micro-LED pixels.
In the oblique direct writing experiment, the S1818 photoresist was also exposed with a 40 µm micro-LED at a current density of 800 A/cm2. The system also used the 50X long working distance microscope objective, and the stage velocity was set to be 10 µm/s, giving a maximum exposure dose of 509.6 mJ/cm2. It can be seen from Figs. 10(a)-(c), the samples were exposed at different tilt angles, ranging from 0° to 30°. Figures 10(d)-(f) show micrographs of lines exposed at sample tilt angles of 0°, 15°, and 30°, corresponding to line widths measured by optical microscopy of 15.1, 14.5, and 12.8 µm, respectively. We found that the linewidths of the exposed lines at different sample tilt angles can be estimated by Eq. (3):
where a is the linewidth of the exposed line at different sample tilt angles, d represents the linewidth of the exposed line when the sample is perpendicular to the microscope objective (i.e. the tilt angle is 0°), and $\theta $ is the sample tilt angle. As shown in Figs. 10(g)-(i), the profilometry was used to test the heights of the exposed lines, which were measured to be 2137, 2090, and 2174 nm, respectively. The exposed lines showed excellent uniformity. The results indicate that the linewidths of the exposed lines decreased by 4.0% and 15.2%, respectively, as the sample tilt angle increased from 0° to 15° and 30°. The linewidth of the exposed line decreases as the sample tilt angle increases, which improves the resolution of the exposure pattern. However, as the sample tilt angle exceeds 30°, the distance between the sample and the microscope objective increases beyond the working distance of the microscope objective, making it impossible to expose a pattern on the sample. The proposed alternative for improving the resolution of the exposure pattern has been confirmed effective. It is worth noting that when multiple micro-LED pixels are used for oblique direct writing, multiple projected spots will be generated on the sample surface, some of which may deviate from the focal plane causing exposure pattern errors. These errors need to be improved by optimizing the depth of focus (DOF) of the system. Furthermore, the lens of the photolithography system inevitably undergoes distortion, which results in exposure pattern errors [27]. This method also has the potential to improve the exposure pattern errors caused by optical distortion of the lens in the photolithography system, thereby enhancing the quality of the exposure pattern.Fig. 10. Oblique direct writing: (a)-(c) schematic diagrams of the tilt setup, (d)-(f) micrographs of exposed lines imaged by an optical microscope with linewidths of 15.1, 14.5, and 12.8 µm, and (g)-(i) profile curves of exposed lines tested by a profilometer with heights of 2137, 2090, and 2174 nm, at sample tilt angles of 0°, 15°, and 30°, respectively.
4. Conclusion
In summary, we have fabricated a 375 nm-emitting micro-LED device and developed a maskless photolithography system based on this device. The system is capable of producing exposure patterns with feature sizes as small as 4.3 µm in photoresist material. The relative intensity distributions of projected spots from the system were analyzed. The study revealed that the relative light intensity in the center area of the projected spot was higher than that in the edge area, leading to an uneven distribution of light intensity and a black border around the exposed dot. Moreover, the effects of exposure time, current density, stage velocity, and current pulse width on the exposure patterns were systematically studied. As exposure time, current density, and current pulse width increase, the exposure dose of the projected spot also increases, resulting in a larger exposure pattern size and lower resolution. Conversely, increasing stage velocity reduces exposure dose and pattern size, leading to higher resolution. In static direct writing, in order to optimize the resolution and quality of the exposure pattern, it is necessary to balance the current density of the micro-LEDs and exposure time. In dynamic direct writing, stage velocity and current pulse width are two key parameters that affect exposure dose and pattern resolution. Additionally, we propose an oblique direct writing method to improve the pattern resolution by adjusting the tilt angle of the sample. This approach does not necessitate a reduction in the size of the micro-LED pixels, making it both practical and cost-effective. The experimental results show that the linewidth of the exposed line decreased by 4.0% when the sample tilt angle was increased from 0° to 15° and by 15.2% when it was increased to 30°, confirming the feasibility of the proposed method. The results demonstrate the properties and advantages of a maskless photolithography system based on micro-LEDs. More real applications of this system will be reported in our future work. The system and method reported in this paper are expected to be applied in various fields such as PCBs, photovoltaics, solar cells, optogenetics, and MEMS.
Funding
National Key Research and Development Program of China (2021YFC2202500, 2021YFC2202503).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. A. Alipour, E. L. Fowler, S. O. R. Moheimani, et al., “Atomic-resolution lithography with an on-chip scanning tunneling microscope,” J. Vac. Sci. Technol. B 40(3), 030603 (2022). [CrossRef]
2. D. Suwandi, R. Aziz, A. Sifa, et al., “Dry film photoresist application to a printed circuit board (PCB) using a maskless photolithography method,” Int. J. Technol. 10(5), 1033–1041 (2019). [CrossRef]
3. Q. Zhang, X.J. Wan, F. Xing, et al., “Solution-processable graphene mesh transparent electrodes for organic solar cells,” Nano Res. 6(7), 478–484 (2013). [CrossRef]
4. G. Kartopu, O. Oklobia, T. Tansel, et al., “A facile photolithography process enabling pinhole-free thin film photovoltaic modules on soda-lime glass,” Sol. Energ. Mat. Sol. C 251, 112112 (2023). [CrossRef]
5. T. Ito and S. Okazaki, “Pushing the limits of lithography,” Nature 406(6799), 1027–1031 (2000). [CrossRef]
6. J. E. E. Baglin, “Ion beam nanoscale fabrication and lithography-a review,” Appl. Surf. Sci. 258(9), 4103–4111 (2012). [CrossRef]
7. M. Mikulics and H. Hardtdegen, “Nano-LED array fabrication suitable for future single photon lithography,” Nanotechnology 26(18), 185302 (2015). [CrossRef]
8. T.R. Groves, D. Pickard, B. Rafferty, et al., “Maskless electron beam lithography: prospects, progress, and challenges,” Microelectron. Eng. 61-62, 285–293 (2002). [CrossRef]
9. J. Sung, B. J. Kang, and J. H. Oh, “Fabrication of high-resolution conductive lines by combining inkjet printing with soft lithography,” Microelectron. Eng. 110, 219–223 (2013). [CrossRef]
10. C. Rensch, S. Hell, M. Vonschickfus, et al., “Laser scanner for direct writing lithography,” Appl. Opt. 28(17), 3754–3758 (1989). [CrossRef]
11. R. M. Guijt and M. C. Breadmore, “Maskless photolithography using UV LEDs,” Lab Chip 8(8), 1402–1404 (2008). [CrossRef]
12. D.H. Dinh, H.L. Chien, and Y.C. Lee, “Maskless lithography based on digital micromirror device (DMD) and double sided microlens and spatial filter array,” Opt. Laser Technol. 113, 407–415 (2019). [CrossRef]
13. D. Elfstrom, B. Guilhabert, J. McKendry, et al., “Mask-less ultraviolet photolithography based on CMOS-driven micro-pixel light emitting diodes,” Opt. Express 17(26), 23522–23529 (2009). [CrossRef]
14. M. Stonehouse, Y.C. Zhang, B. Guilhabert, et al., “Digital illumination in microscale direct-writing photolithography: challenges and trade-offs,” Proc. British Irish Conf. Opt. Photon. 1, 1–4 (2019). [CrossRef]
15. B. Guilhabert, D. Massoubre, E. Richardson, et al., “Sub-micron lithography using InGaN micro-LEDs: mask-free fabrication of LED arrays,” IEEE Photon. Technol. Lett. 24(24), 2221–2224 (2012). [CrossRef]
16. Z. Wang, X.Y. Shan, S.J. Zhu, et al., “Size-dependent sidewall defect effect of GaN blue micro-LEDs by photoluminescence and fluorescence lifetime imaging,” Opt. Lett. 48(18), 4845–4848 (2023). [CrossRef]
17. P.F. Tian, X.Y. Liu, S.Y. Yi, et al., “High-speed underwater optical wireless communication using a blue GaN-based micro-LED,” Opt. Express 25(2), 1193–1201 (2017). [CrossRef]
18. X.Y. Liu, L.Y. Tao, S.L. Mei, et al., “White-light GaN-µLEDs employing green/red perovskite quantum dots as color converters for visible light communication,” Nanomaterials 12(4), 627 (2022). [CrossRef]
19. Z. Wang, S.J. Zhu, X.Y. Shan, et al., “Red, green and blue InGaN micro-LEDs for display application: temperature and current density effects,” Opt. Express 30(20), 36403–36413 (2022). [CrossRef]
20. L. W. Liebmann, “Resolution enhancement techniques in optical lithography: it's not just a mask problem,” Proc. SPIE 4409, 23–32 (2001). [CrossRef]
21. J. Chung, N. R. Bieri, S. Ko, et al., “In-tandem deposition and sintering of printed gold nanoparticle inks induced by continuous Gaussian laser irradiation,” Appl. Phys. A 79(4-6), 1259–1261 (2004). [CrossRef]
22. T. Schwarz-Selinger, D.G. Cahill, S.C. Chen, et al., “Micron-scale modifications of Si surface morphology by pulsed-laser texturing,” Phys. Rev. B 64(15), 155323 (2001). [CrossRef]
23. K. Nakamura, J. Miyazu, M. Sasaura, et al., “Wide-angle, low-voltage electro-optic beam deflection based on space-charge-controlled mode of electrical conduction in KTa1− xNbxO3,” Appl. Phys. Lett. 89(13), 131115 (2006). [CrossRef]
24. Z. G. Ling, K. Lian, and L.K. Jian, “Improved patterning quality of SU-8 microstructures by optimizing the exposure parameters,” Proc. SPIE 3999, 1019–1027 (2000). [CrossRef]
25. C.W. Jeon, E. Gu, and M.D. Dawson, “Mask-free photolithographic exposure using a matrix-addressable micropixellated AlInGaN ultraviolet light-emitting diode,” Appl. Phys. Lett. 86(22), 221105 (2005). [CrossRef]
26. M. S. Wong, D. Hwang, A. I. Alhassan, et al., “High efficiency of III-nitride micro-light-emitting diodes by sidewall passivation using atomic layer deposition,” Opt. Express 26(16), 21324–21331 (2018). [CrossRef]
27. H.L. Chien, Y.H. Chiu, and Y.C. Lee, “Maskless lithography based on oblique scanning of point array with digital distortion correction,” Opt. Lasers Eng. 136, 106313 (2021). [CrossRef]
