Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Understanding the Sidewall Passivation Effects in AlGaInP/GaInP Micro-LED

Understanding the Sidewall Passivation Effects in AlGaInP/GaInP Micro-LED The passivation effects of sulfur treatment and Al O passivation for AlGaInP/GaInP red micro‑light ‑ emitting‑ diodes 2 3 (LEDs) were investigated in terms of the external quantum efficiency (EQE) and the current density showing the peak EQE (J ). We systematically compared the electrical and optical characteristics of the micro‑LEDs with and with‑ EQE, peak out passivation according to various sizes. Interestingly, our investigation indicated that simple electrical characteris‑ tics such as current density–voltage property are difficult to precisely reflect the minor change in electrical properties due to passivation when the device has the inherently low leakage current. Whereas the EQE was enhanced by 20% and J was largely shifted to a lower current density region at the LED with a size of 15 × 15 μm . To examine the EQE, peak passivation effects, we carefully analyzed the EQE and J with the ABC recombination model, and established EQE, peak the methodology to investigate the influence of a sidewall in micro ‑LEDs. As a result, we extracted the surface recom‑ bination velocity regarding the surface passivation, showing a nearly 14% reduction with the passivation. Keywords: Micro‑LED, Passivation, AlGaInP/GaInP MQW etching (ICP-RIE), play a critical role as a Shockley– Introduction Read–Hall (SRH) recombination center. It results in the Inorganic micro-light-emitting-diodes (LEDs) are in the degradation of the EQE value especially for micro-LEDs spotlight for ultra-high resolution display devices due to because the smaller LEDs have a large surface-to-volume its high performance such as high brightness, scalabil- ratio as the device size decreases. Also it could cause the ity, and contrast ratio [1, 2]. In order to implement the current density showing the maximum EQE (J ) energy-efficient micro-LED display, high external quan - EQE, peak shift to a higher current density region [6, 7]. tum efficiency (EQE) should be achieved with a scaled Therefore, to suppress the sidewall induced perfor - pixel smaller than 50 × 50  μm which is a 1/100 size of mance degradations, the various passivation strate- a conventional LED [3]. Furthermore, a maximum EQE gies have been studied for inorganic micro-LEDs. For should be positioned at the relatively low injection cur- instance, the KOH treatment followed by an atomic layer rent regime, which could lead to a strong benefit of a deposition system (ALD) has been reported to passivate power consumption for micro-display compared with the InGaN micro-LED. With passivation, the light emis- the LED for general lighting [4, 5]. However, the sidewall sion uniformity was improved, the EQE reduction by defects, which are mainly caused by the mesa formation shrinking micro-LED dimensions was effectively reduced process such as inductively-coupled plasma reactive ion [8]. In the case of the AlGaInP/GaInP multi-quantum well (MQW) based red micro-LEDs, it is known to have a larger surface recombination velocity (SRV) than InGaN/ *Correspondence: gdmgdm@kaist.ac.kr; shkim.ee@kaist.ac.kr Dae‑Myeong Geum and Sanghyeon Kim are co ‑ corresponding authors GaN MQW, resulting in significant performance degra - and they contributed equally. dation with a pixel size scaling [9]. However, there are School of Electrical Engineering, Korea Advanced Institute of Science still few researches on the AlGaInP/GaInP passivation, and Technology (KAIST ), Daejeon 34141, Republic of Korea Infromation and Electronics Research Institute, Korea Advanced Institute although red LED is one of the important building blocks of Science and Technology (KAIST ), Daejeon 34141, Republic of Korea © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Park et al. Nanoscale Research Letters (2022) 17:29 Page 2 of 9 for fabricating future full-color micro-displays [10, 11]. for 10  min to reduce the contact resistance of ohmic Furthermore, most of the reported micro-LEDs showed contacts. a high leakage current, indicating the fabrication process After the standard device fabrication, we carried out has not much been optimized than InGaN/GaN LED. the surface passivation. The micro-LEDs were dipped On the other hand, the investigation of EQE relative to in HCl: DI (1:5) solution for 30  s to remove the native injection current density has not been conducted yet in oxide of the sidewall. Then, the sample was dipped in AlGaInP/GaInP MQW micro-LEDs because a low cur- (NH ) S :DI (1:10) solution for 10  min and was subse- 4 2 x rent injection regime would not be critically considered quently loaded into the atomic layer deposition (ALD) for lighting applications. Therefore, to fabricate the highly chamber to minimize the naturally formed native oxide. efficient micro-displays, the exploration of the fabrication Finally, after depositing the 10  nm Al O passivation 2 3 technology of AlGaInP/GaInP LED including pixel for- layer, the passivated micro-LEDs were analyzed com- mation process as well as surface passivation should be paratively with non-passivated micro-LED. The elec - carried out. In addition, the investigation of EQE relative trical characteristics were measured by Keithley 4200. to injection current density including J should be For measuring optical characteristics, we firstly used EQE, peak conducted to develop the red-pixels of display applica- the integrated sphere system. However, when using the tion and understand the effect of sidewall degradation. integrated sphere, inevitable photon loss occurred since In the present work, we fabricated the AlGaInP/ the micro-LED is located a few inches away from the GaInP red micro-LEDs and systematically investigated inner surface of the sphere. Therefore, to measure the the EQE with various device sizes from 15 × 15  μm to lower current density region, we used the photodetec- 80 × 80  μm . Furthermore, to investigate the passivation tor located directly above the micro-LED and calibrated effect for AlGaInP/GaInP micro-LEDs, we conducted the the measured photocurrent to light output power (W) by conventional surface passivation with sulfur treatment referring to the integrated sphere data (Additional file  1: followed by Al O deposition [12–14]. Furthermore, the Figure S1). 2 3 effect of the passivation was carefully examined in terms of the EQE enhancement and J shift through opti- Results and Discussion EQE, peak cal and electrical characterizations. Then, to quantitively Electrical and Optical Characterization of the Fabricated extract the SRV values, by fitting the measured EQE data Micro‑LEDs with the ABC model of recombinations, the SRH recom- The electrical characteristics of the AlGaInP/GaInP bination constants were well analyzed with respect to the micro-LED without passivation were shown in Fig.  1b micro-LED sizes. depending on the device sizes from 15 × 15  μm to 80 × 80  μm . At the reverse bias region, the dark current Methods densities were measured nearby the measurement floor, Fabrication Process of Micro‑LEDs while it was clearly noted that there is a low leakage cur- The epitaxial layers were grown by metal–organic chemi - rent that originates from parasitic current paths such as cal vapor deposition (MOCVD) on a 4-inch semi-insu- a sidewall [15]. Comparing the 15 × 15  μm device char- lating GaAs substrate. The micro-LED structure based acteristics with the same sized device of Oh et al. our IV on AlGaInP/GaInP MQW structure is shown in Fig.  1a. plot shows low leakage current and dark current density The micro-LED fabrication process was started with the level. It suggested that there was negligible degradation standard cleaning process with acetone, methanol, and through various fabrication processes. Also, the system- deionized (DI) water. Sequentially, the mesa isolation atic current density increase over 2 V bias is found as the was conducted to define the pixel by ICP-RIE. We etched device size decreases. In Fig.  1c, the J–L (Current den- the AlGaInP/GaInP epilayers with C l and Ar gas flow sity–Light output power) characteristics of 20 × 20  μm in an ICP-RIE system. In this step, the etching was car- are shown by comparing the same size device of Wong ried out until the middle of the n-GaInP layer. Then, the et  al. [10]. Compared to the device of Wong et  al. the samples were etched to completely remove the remain- 20  μm micro-LED of this paper showed 2.32  μW, which + 2 ing n-GaInP layers and expose the n -GaAs contact is 46 times higher at 20 A/cm . And 6.0 μW which is 54 layer using the H PO : HCl (3:1) solution. For preventing times at 40 A/cm . They passivated the devices by depos - 3 4 electrical short problem of p-contact metal, Al O layer iting Al O followed by nitrogen plasma surface treat- 2 3 2 3 was patially defined for p-conact region. After that, Cr/ ment. Even comparing with the passivated micro-LED, Au (25/75  nm) was deposited for p-type contact metal our micro-LED showed 8 times higher at 20  A/cm and to GaP by the electron beam (e-beam) evaporator. Also, 8 times higher at 40  A/cm . As a result, our micro-LED Pd/Ge/Au (20/40/100  nm) was deposited on the n - device showed a much larger light emission. The inset GaAs layer. Finally, the sample was annealed at 200  °C of Fig.  1d shows the microscope image with 100  A/cm P ark et al. Nanoscale Research Letters (2022) 17:29 Page 3 of 9 Fig. 1 a The process flow of AlGaInP/GaInP MQW LED with passivation. b The current density–voltage (J–V) characteristic of AlGaInP/GaInP MQW LED depending on device sizes. c The current density–light output power (J–L) characteristics of 20 × 20 μm micro‑LED comparing with the same size devices of Ref [10] d Electroluminescence of AlGaInP/GaInP MQW LED with different current density. The inset in (d) is the microscope image of 2 2 20 × 20 μm AlGaInP/GaInP MQW LED at 100 A/cm current density injected to micro-LEDs. When the cur- measurements were conducted for non-passivated and 2 2 rent of 100 A/cm is driven, the red light emission to the passivated micro-LEDs with 15 × 15 μm . In Fig. 2a, there top surface of the 20 × 20  μm device is clearly shown. were almost the same current density–voltage curves at As shown in Fig.  1d, the electroluminescence (EL) spec- the whole bias range. For other device sizes, it was noted tra of 15 × 15  μm LEDs are shown with the peak wave- that there were only negligible changes (not shown here). length 628  nm with current densities in ranges from 53 Compared with other reports having relatively high leak- to 400  A/cm . Resulting full-width at half maximum age current, despite passivation, the reduction of leak- (FWHM) value at 100  A/cm is 14  nm, which is simi- age current was not noticeable in our micro-LEDs [7, lar to previous reported LEDs [16]. Furthermore, it was 10, 11]. Specifically, in Figs.  1b and 2a in our study, the −9 found that the light intensity increases with an increase leakage current of reverse bias was saturated from 10 −7 2 of the injection current density, and without the severe to 10  A/cm in all device dimensions. The leakage cur - peak wavelength change. From these results, the fabri- rent showed a very low level and the saturation behavior cated micro-LEDs are well-functioned with our fabrica- was very stable with increasing reverse bias. Additionally, tion process. in Figs. 1b and 2a, the J–V characteristics were shown in To examine the passivation effects on the electri - the log scale. Therefore, we can check there is no addi - cal characteristics, current density–voltage (J–V) tional leakage current path nearby the threshold voltage Park et al. Nanoscale Research Letters (2022) 17:29 Page 4 of 9 Fig. 2 a J–V curves of non‑passivated and passivated 15 × 15 μm LED. b Diode ideality factor depending on device sizes. c The EQE comparison of 2 2 AlGaInP/GaInP LEDs which have the size of 15 × 15 μm and 80 × 80 μm with and without passivation. d The enhancement ratio of maximum EQE with passivation depending on the device sizes −1 at forward bias. These results strongly suggest that our q ∂ ln I n = (1) devices inherently showed low leakage current level at kT ∂V both reverse and forward bias, thereby, the negligible variation for electrical properties was observed even after where n, q, k, T, I, and V are the ideality factor, the ele- the surface passivation rather than other reports. It could mentary charge, the Boltzman constant, the absolute be attributed to the inherently low leakage current with temperature, current, and the applied voltage, respec- well-defined fabrication process, which could lead to very tively. Figure  2b illustrates the ideality factors of non- small deviation of leakage currents. passivated and passivated micro-LEDs depending on 2 2 For the in-depth analysis, the diode ideality factor was the device sizes from 15 × 15  μm to 100 × 100  μm . The calculated by Eq. (1) value of the 425 × 425  μm LED without passivation was also added for a large-size LED. Referring to the conven- tional interpretation of ideality factor in micro-LEDs, the P ark et al. Nanoscale Research Letters (2022) 17:29 Page 5 of 9 ideality factor near unity means a radiative recombina- Figure 2c shows the normalized EQEs with and without 2 2 tion dominance, while the ideality factor of 2 indicates passivation for 15 × 15 μm and 80 × 80 μm micro-LEDs. SRH recombination dominance through trap states [17]. The values of EQE are normalized by the maximum val - As depicted in Fig.  2b, the value of the ideality factor of ues of passivated micro-LEDs for each size. The enhance - the non-passivated device is from 1.82 (425 × 425 μm ) to ment of the peak EQE values was clearly observed in the 1.93 (80 × 80 μm ). The severe degradation of ideality fac - optical measurement unlike the electrical characteristics. tors depending on mesa sizes is not found compared to Even though the electrical characteristics didn’t show previous reports. Additionally, ideality factors of the pas- definite difference because of the low leakage current sivated micro-LEDs are from 1.89 (80 × 80  μm ) to 1.94 level, the optical performance showed clear improvement (15 × 15  μm ), which are very similar to non-passivated by utilizing the passivation process. LEDs. When compared to other reports, the notice- To compare the effects of surface passivation depend - able increase of ideality factors was not observed with a ing on a different size, the enhancement ratio is defined reduction of micro-LED sizes and passivation. Although by Eq. (2). Max. EQE(passivated) − Max. EQE(non passivated) (2) Enhancement ratio = Max. EQE(non passivated) The surface passivation increased the maximum EQE the ideality factor has been used as a good indirect indi- of 15 × 15  μm micro-LED as 19.8% and the maximum cation for electrical characteristics of LED, it was not EQE of 80 × 80  μm as a 2.4%. Because of the higher sur- so effective to analyze the fabricated LEDs in this work. face-to-volume ratio, 15 × 15  μm was more affected by The difference of ideality factors depending on the device surface recombination of sidewall defects, so the passiva- sizes are subtle like within 2 decimal places, thus it was tion effect was larger than 80 × 80  μm . Besides, calcu- quite difficult to evaluate the sidewall degradation or pas - lated enhancement ratios for the different sizes is shown sivation effects from these values. Because our micro- in Fig. 2d. The 60 μm and 80 μm micro-LEDs showed an LED devices showed an inherently low leakage current, enhancement ratio of less than 5%, but the device pitch which is close to measurement limits, the minor change smaller than 40  μm had a higher value over 10%. The in electrical characteristics with or without passivation enhancement ratio starts to increase rapidly under the couldn’t be measured. device size of 40 μm. The large enhancement of EQEs in In the fabrication process, the etching process using the smaller size micro-LEDs strongly suggests that our BOE and HCl was included. The buffered oxide etchant passivation could suppress the surface recombination (BOE) and HCl have been reported as a good remover for the AlGaInP/GaInP micro-LEDs smaller than 40  μm of the native oxide at the surface of the semiconduc- size. On the other hand, the EQE enhancement might be tor. Thus, the electrical characteristics may look better attributed to the increased LEE. Due to the lower refrac- immediately after immersion in the solutions. However, tive index of Al O, Al O deposition decreases the dif- 2 3 2 3 if there is no additional treatment like sulfur immer- ference of refractive index between LED epilayers and air. sion, the sidewall will be gradually re-oxidized after Therefore, less amount of photons would be guided in the enough time passes [18]. Because the micro-LEDs with- epi-layers due to less total internal reflection. The LEE out passivation in the manuscript are measured after enhancement effect and the passivation effect seem to be being exposed to air quite a few hours later, it doesn’t mixed in the results. To clarify this issue, we simulated seem the wet treatments critically change the electrical 40  μm micro-red-LED (GaP/Al(Ga)InP/MQW/Al(Ga) measurement. Especially, HCl has been widely utilized InP/GaAs substrate) by finite-difference time-domain as an etchant for AlGaInP epilayers or GaInP etch stop (FDTD) simulation of LUMERICAL. The dipole source layer combining with H PO when fabricating the red- 3 4 with the wavelength 625  nm is positioned in the center LEDs [19–21]. However, those studies have not shown of MQWs and the transmittance towards the top and special electrical characteristics or high external quan- sidewall is calculated when 10  nm Al O is deposited or 2 3 tum efficiency like this paper. This indirectly indicates not. As a result, the device with Al O had a 1% decrease 2 3 that the usage of the chemicals such as H PO , HCl, 3 4 in top transmittance and no difference at the sidewall, BOE is not the main factor influencing the micro- which means if the thickness is not optimized, there is LED performance. Also, all wet processes were equally no LEE enhancement. Therefore, the EQE improvements addressed to each sample, thus, the impact of the passi- of this work are not because of enhanced LEE by Al O 2 3 vation steps was fairly compared and the improvement by passivation steps is reasonable. Park et al. Nanoscale Research Letters (2022) 17:29 Page 6 of 9 Fig. 3 a Size‑ dependent EQE of AlGaInP/GaInP micro‑LED. b Peak current density of LED devices (symbol) and the current density range 1% apart from peak current density (line) with (left) and without passivation (right). c The EQE plot fitting based on the ABC model. d Size ‑ dependent SRH parameters were used in fitting the EQE curves with and without passivation. The surface recombination velocity is extracted from the slope deposition but because of the reduced SRH recombina- The Extraction of a Surface Recombination Velocity tion by passivation process. for Micro‑LEDs Although the EQE enhancements were observed, To quantitatively confirm the sidewall contribution, it is through a simple optical characterization, we could necessary to evaluate the passivation effect not just by the relatively evaluate the effects of the passivation such simple electrical characteristics but by J analysis EQE, peak as enhancement ratios. Thus, to compare the sidewall with optical measurement. J can be a useful ana- EQE, peak contribution of micro-LEDs, size-dependent electrical lytical tool to analyze the impact of SRH recombination. characterization, other than J–V curves or the ideality Because J is proportional to A/C , in which A EQE, peak factor, and optical characterization should be simulta- is the SRH recombination coefficient and C is the auger neously considered. recombination coefficient respectively [22]. In Fig.  3a, the EQE of micro-LED without passivation is shown as a function of the current density with various P ark et al. Nanoscale Research Letters (2022) 17:29 Page 7 of 9 sizes. In the literature, when the sidewall degradation has η Bn inj η = η η , η = , (3) ext lee int int a significant impact on micro-LED, the smaller devices 2 3 An + Bn + Cn have lower EQE because of higher sidewall degradation [7]. However, in Fig.  3a the peak EQE is interestingly 2 3 J = qw An + Bn + Cn increased when the size is varied from 80 to 40 μm, and (4) the peak EQE is decreased with a decrease the size from 40 to 15  μm. These results indicate a relatively lower impact of sidewall degradation than other reports due to A = A + v (5) 0 s the well-designed fabrication process and the impact of increased light extraction efficiency (LEE). Here, η , η , η , η , and n are the external quan- ext lee int inj As the mesa size is decreased, the EQE curve shifts to tum efficiency, the light extraction efficiency (LEE), the a higher current density region, and J becomes internal quantum efficiency, the injection efficiency of EQE, peak higher. The 15 × 15 μm device showed J at 444 A/ the LED, and the carrier concentration, respectively. EQE, peak 2 2 cm , whereas relatively large 80 × 80  μm device showed The ABC model assumes that the A, B, C constants and 27  A/cm . It shows a very significant feature of the deg - η of the LED chip are independent of the carrier con- lee radation associated with the size scaling of the LED, centration and operating current. The η is affected lee indicating that the analysis of J would be a key to by factors that can influence the optical path of emit - EQE, peak capture the impact of the surface state at the sidewall. ted photons such as device structures and layer thick- In Fig. 3b, the bar plot shows the range of current den- ness. Those factors don’t change when the miro-LEDs sity which has the EQE value within 1% deviation at the are operated with different current injection. Therefore, maximum point. The diamond symbol points to J . the LEE can be considered as constant. Also, the η is EQE, peak inj Dry etched sidewall surface could have a non-uniform deserved as 100% at the low current density region in surface condition in terms of defect density (damage). the MQW [23]. Therefore, considering constant η and lee Therefore, before the annihilation of the defective region η , the experimental normalized EQE plot can be fitted inj by the surface passivation, this non-uniformity might by the normalized η in Eq.  (3). The current density of int contribute to the variation of the J , as we meas- the ABC model is calculated by Eq.  (4). J and w are the EQE, peak ured. Therefore, we indicated J value with error current density and the width of the quantum well. For EQE, peak bar. The black and red plots represent the data without the w value, 6 nm was used referring to the thickness of passivation and with passivation for each. Comparing the the quantum well of the epitaxial structure used in this two plots, the J shift was reduced with the passiva- study. The carrier concentrations (n) were swept from EQE, peak 15 19 −3 tion process. J of the 15 × 15  μm device without 10 to 10  cm to calculate EQE as a function of the EQE, peak the surface passivation ranges from 347 to 666  A/cm . current density on a log scale. Based on those equations, With the surface passivation, the range is lowered from the normalized measured EQE was fitted by the calcu - 2 2 302 to 444  A/cm . On the other hand, the 40 × 40  μm lated internal quantum efficiency of the ABC recom - device is relatively less affected than a 15 × 15 μm device bination model. As the most governing parameter, we by sidewall recombination. Furthermore, the 60 × 60 μm varied the values of the A value to investigate the impact and 80 × 80  μm size devices showed a slightly lowered of the surface recombination. On the other hand, we −10 3 −1 peak current density, but the effect was negligibly small. fixed the B and C coefficient as 1.5 × 10  cm  s and −29 6 −1 Therefore, we can conclude that the sidewall contribu - 4 × 10  cm  s since those parameters have minor tion of AlGaInP/GaInP micro-LED is very critical under changes on fitting results of the target current density the size of 40 × 40  μm , which is the dimension of our range (Additional file  2: Figure S1)  [24, 25]. In Fig.  3c, interest for the display to achieve high EQE and maintain the normalized EQE of 20 × 20  μm is fitted by different 7 −1 7 −1 peak current density. Also, it can be effectively reduced A values from 4.2 × 10  s to 6.2 × 10  s . Increasing A through wet treatment followed by depositing the Al O parameter moves to the J to higher current den- 2 3 peak, EQE passivation layer. sity, reflecting the increased SRH recombination at the To deeply investigate the origin, analyzing the effi - sidewall. Therefore, we chose the reasonable A values of ciency behavior by fitting through the ABC model is a good fitting results which induce less than 5% fitting dif - useful strategy. The A, B, and C constant represents the ference between experimental data and calculated result. SRH non-radiative recombination, the radiative recom- The fitting difference is defined as the ratio (%) of the dif - bination, and the Auger recombination for each. In the ference between normalized EQE and normalized IQE ABC model, the external quantum efficiency is expressed to normalized EQE. The best fitting result is plotted by 7 −1 as follows [22]. 5.20 × 10  s of A parameter in Fig.  3c. Our target fit - ting region is the current density around J and EQE, peak, Park et al. Nanoscale Research Letters (2022) 17:29 Page 8 of 9 under J . This region is where SRH recombination amount of the trap along with the energy level will fur- EQE, peak (A parameter) has a dominant impact on IQE calcula- ther improve the performance. tion and the magnitude of sidewall degradation can be shown. At high current region, because there are other Conclusion hindrances such as current crowding effect or high series In conclusion, we fabricated the AlGaInP/GaInP micro- resistance, it is hard to obtain perfectly matched fitting LEDs and investigated the electrical and optical charac- results. Therefore, we fitting process was focused on 2 teristics of the devices with the sizes from 15 × 15  μm the low current density region including J where 2 EQE, peak to 80 × 80  μm . In this process, the specific EQE value LEDs show the inherent performances without additional of the Red LED pixel was investigated to a low current effects. The yellow and blue shadowing region represents density range and this data will be an important refer- which recombination mechanism is dominant. ence aimed at using the AlGaInP/GaInP red LED as a The constant A which is the SRH recombination coef - unit pixel micro-LED display. Also, the investigation on ficient can be re-written with the bulk SRH coefficient the effect of surface passivation with sulfur treatment A the surface recombination velocity (SRV) v , p erim- 0, followed by Al O deposition was conducted. Since our 2 3 eter P and area S in Eq.  (5). The bulk term is A which micro-LEDs showed the inherently low leakage current describes that the SRH recombination at the bulk epi-lay- before passivation with well-arranged fabrication pro- ers, and the perimeter term is v · which describes the cess, there was no notable electrical difference with and SRH recombination at the device surface. To focus on the without passivation. On the other hand, the clear EQE quantitative value of v , we utilized the A, B, C recombi- improvement and lowered J was observed par- EQE, peak nation coefficients as keys to connect the experimental 2 ticularly in micro-LED with a size under 40 × 40  μm . data and SRH recombination coefficient. Therefore we This motivated us to suggest a new analysis method to can extract the SRV by calculating the slope of A param- provide an insight into the surface state of the LEDs. eter versus P/S ratio depending on the device size by Therefore, we first introduced the novel analysis meth - Eq.  (5). In Fig.  3d, the SRH recombination constant (A) odology with J to evaluate the sidewall effect in EQE, peak of with and without passivation are shown with the error micro-LED. Additionally, we fitted the experimental bars. The median values of each range showed the best- EQE with the ABC recombination model and extracted matched fitting results and they were used to calculate the SRV from the size-dependent SRH constant (A), SRV. In the case of passivation, the value of 27,445 cm/s and demonstrated a 14% improvement with the passi- was 15% lower than that of 32,439 cm/s without passiva- vation. The evaluation process shown in this paper can tion by extracting the slopes. Although the extracted SRV be used as a straightforward, but simple guideline to values were not be decreased significantly, it is worth analyze the sidewall-related performance degradation noting that the change of J shift depending on EQE, peak or improvement of micro-LED. Consequently, to real- the various micro-LED size allows us to extract the SRV ize high EQE red micro-LED, further studies that can values. Although the extracted SRV values were not be remove the trap sites at midgap and/or near the valence decreased significantly, it is worth noting that the change band should be carried out in the future. of J shift depending on the various micro-LED EQE, peak size allows us to extract the SRV values. Abbreviations Consequently, even if we achieved the peak EQE Micro‑LED: Micro ‑light ‑ emitting‑ diode; EQE: External quantum efficiency; SRV: enhancement, the lower J , and the SRV reduc- EQE, peak Surface recombination velocity; ICP‑RIE: Inductively‑ coupled plasma reactive tion, the enhancement of device performance was not ion etching; SRH: Shockley–Read–Hall; ALD: Atomic layer deposition; MQW: Multi‑ quantum well; MOCVD: Metal–organic chemical vapor deposition; so dramatic, indicating further study should be care- DI: Deionized; EL: Electroluminescence; FWHM: Full with at half maximum; fully designed and carried out to maximize the perfor- BOE: Buffered oxide etchant; FDTD: Finite ‑ difference time ‑ domain; LEE: Light mance of scaled LEDs. The possible reason for limited extraction efficiency. enhancement in this work is the incomplete removal of the defects at the sidewall. Since the sulfur treatment and Supplementary Information the Al O passivation are known to remove the native The online version contains supplementary material available at https:// doi. 2 3 org/ 10. 1186/ s11671‑ 022‑ 03669‑5. oxide of GaAs and reduce the surface density states near the conduction band edge [28], thereby, the surface Additional file 1: Supplementary information for optical measurement trap states at the midgap and/or near the valence band for low current density: The photodetector measurement. can still be remained and limit the device performance Additional file 2: Supplementary information for selection fitting enhancement even with the passivation. This suggests parameter. that further passivation study with a consideration of the P ark et al. Nanoscale Research Letters (2022) 17:29 Page 9 of 9 Authors’ Contributions In Ga As and insulator by H annealing with Pt gate electrode. Appl 0.53 0.47 2 PJH, BWJ, were involved in experiments. PJH, BWJ, GDM, and SHK participated Phys Lett 115:143502 in analysis and discussions. PJH, GDM, and SHK drafted the manuscript. All 14. Kim S‑H, Yokoyama M, Nakane R, Ichikawa O, Osada T, Hata M, Takenaka authors read and approved the final manuscript. M, Takagi S (2014) High performance tri‑ gate extremely thin‑body InAs‑ On‑Insulator MOSFETs with high short channel effect immunity and Vth Funding tunability. IEEE Trans Electron Devices 61(5):1354–1360 This work was supported by Samsung Research Funding & Incubation Center 15. Cchubert EF (2006) Ligh‑ emitting diodes, 2nd edn. Cambridge University of Samsung Electronics under Project number SRFC‑ TB2003‑02. Press, Cambridge, pp 65–66 16. Zhao Y, Liang J, Zeng Q, Li Y, Li P, Fan K, Sun W, Lv J, Qin Y, Wang Q, Tao J, Availability of Data and Materials Wang W (2021) 2000 PPI silicon‑based AlGaInP red micro ‑LED arrays fabri‑ The datasets used and analysed during the current study are available from cated via wafer bonding and epilayer lift‑ off. Opt Express 29:20217–20228 the corresponding author on reasonable request. 17. Sah CT (1991) Fundamentals of solid‑state electronics. World Scientific, Singapore, pp 438–443 18. Petrovykh DY, Yang MJ, Whitman LJ (2003) Chemical and electronic Declarations properties of sulfur‑passivated InAs surfaces. Surf Sci 523:231–240 19. Tseng MC, Chen CL, Lai NK, Chen SI, Hsu TC, Peng YR, Horng RH (2014) Competing interests P‑side ‑up thin‑film AlGaInP ‑based light ‑ emitting diodes with direct The authors declare that they have no competing interests. ohmic contact of an ITO layer with a GaP window layer. Opt Expres 22:A1862–A1867 Received: 23 December 2021 Accepted: 15 February 2022 20. Han S, Xu C, Li H, Liu S, Xu H, Zhu Y, Fang A, Wang X (2021) AlGaInP‑based Micro‑LED array with enhanced optoelectrical properties. Opt Mater 114:110860 21. Chilukuri K, Mori MJ, Dohrman CL, Fitzgerald EA (2006) Monolithic CMOS‑ compatible AlGaInP visible LED arrays on silicon on lattice‑ engineered References substrates (SOLES). Semicond Sci Technol 22(2):29–34 1. Kim SH, Kim SK, Shim JP, Geum DM, Ju GW, Kim HS, Lim HJ, Lim HR, Han 22. Smith JM, Ley R, Wong MS, Baek YH, Kang JH, Kim CH, Gordon MJ, Naka‑ JH, Lee SB, Kim HS, Bidenko P, Kang CM, Lee DS, Song JD, Choi WJ, Kim HJ mura S, Speck JS, Denbaars SP (2020) Comparison of size‑ dependent (2018) Heterogeneous integration toward a monolithic 3‑D chip enabled characteristics of blue and green InGaN microLEDs down to 1um in by III–V and Ge materials. IEEE J Electron Devices Soc 6:579–587 diameter. Appl Phys Lett 116:071102 2. Geum DM, Kim SK, Kang CM, Moon SH, Kyhm JH, Han JH, Lee DS, Kim SH 23. Zhao HP, Liu GG, Zhang J, Arif RA, Tansu N (2013) Analysis of internal (2019) Strategy toward the fabrication of ultrahigh‑resolution micro ‑LED quantum efficiency and current injection efficiency in III‑nitride light ‑ displays by bonding‑interface ‑ engineered vertical stacking and surface emitting diodes. J Display Technol 9:212–225 passivation. Nanoscale 11:23139–23148 24. Levinshtein M, Rumyantsev S, Shur M (1999) Handbook series on semi‑ 3. Wu T, Sher C‑ W, Lin Y, Lee C‑F, Liang SJ, Lu YJ, Chen S‑ WH, Guo WJ, Kuo conductor parameters. Tenary and quaternary III–V compounds, vol 2. H‑ C, Chen Z (2018) Mini‑LED and micro ‑LED: promising candidates for World Scientific, Singapore the next generation display technology. Appl Sci 8(9):1557 25. Brendel M, Kruse A, Jönen H, Hoffmann L, Bremers H, Rossow U, 4. Day J, Li J, Lie DYC, Bradford C, Lin JY, Jiang HX (2011) III‑Nitride full‑scale Hangleiter A (2011) Auger recombination in GaInN/GaN quantum well high‑resolution microdisplays. Appl Phys Lett 99:031116 laser structures. Appl Phys Lett 99:031106 5. Chong WC, Cho WK, Liu ZJ, Wang CH, Lau KM (2014) 1700 Pixels per 26. Olivier F, Daami A, Licitra C, Templier F (2017) Shockley–Read–Hall and inch (PPI) passive‑matrix micro ‑LED Display powered by ASIC. In: IEEE Auger non‑radiative recombination in GaN based LEDs: a size effect compound semiconductor integrated circuit symposium (CSICS). pp 1–4 study. Appl Phys Lett 111:022104 6. Tian P, McKendry JJD, Gong Z, Guilhabert B, Watson IM, Gu E, Chen Z, 27. Verzellesi G, Saguatti D, Meneghini M, Bertazzi F, Goano M, Meneghesso Zhang G, Dawson MD (2012) Size‑ dependent efficiency and efficiency G, Zanoni E (2013) Efficiency droop in InGaN/GaN blue light ‑ emitting droop of blue InGaN micro‑light emitting diodes. Appl Phys Lett diodes: physical mechanisms and remedies. Appl Phys Lett 114:071101 101:231110 28. Zhou L, Bo B, Yan XZ, Wang C, Chi YD, Yang XT (2018) Brief review of 7. Oh JT, Lee SY, Moon Y T, Moon JH, Park SW, Hong KY, Song KY, Oh CH, surface passivation on III–V semiconductor. Curr Comput‑Aided Drug Des Shim JI, Jeong HH, Song JO, Amano H, Seong TY (2018) Light output 8(5):226 performance of red AlGaInP‑based light emitting diodes with different chip geometries and structures. Opt Express 26:11194–11200 Publisher’s Note 8. Wong MS, Lee CM, Myers DJ, Hwang D, Kearns JA, Li T, Speck JS, Naka‑ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ mura S, Denbaars SP (2019) Size‑independent peak efficiency of III‑nitride lished maps and institutional affiliations. micro‑light ‑ emitting‑ diodes using chemical treatment and sidewall passivation. Appl Phys Express 12:097004 9. Bulashevich KA, Karpo SY (2016) Impact of surface recombination on efficiency of III‑nitride light ‑ emitting diodes. Phys Status Solidi RRL 10(6):480–484 10. Wong MS, Kearns JA, Lee CM, Smith JM, Lynsky C, Lheureux G, Choi HS, Kim JW, Hon Kim C, Nakamura S, Speck JS, DenBaars SP (2020) Improved performance of AlGaInP red micro‑light ‑ emitting diodes with sidewall treatments. Opt Express 28:5787–5793 11. Huang HH, Huang SK, Tsai YL, Wang SW, Lee YY, Weng SY, Kuo HC, Lin CC (2020) Investigation on reliability of red micro‑light emitting diodes with atomic layer deposition passivation layers. Opt Express 28(25):38184–38195 12. Suzuki R, Taoka N, Yokoyama M, Lee S, Kim SH, Hoshii T, Yasuda T, Jevasu‑ wan W, Maeda T, Ichikawa O, Fukuhara N, Hata M, Takenaka M, Takagi S (2012) 1‑nm‑ capacitance‑ equivalent‑thickness HfO /Al O /InGaAs metal‑ 2 2 3 oxide‑semiconductor structure with low interface trap density and low gate leakage current density. Appl Phys Lett 100:132906 13. Kim SK, Geum D‑M, Lim H‑R, Kim HS, Han J‑H, Hwang DK, Song JD, Kim H‑ J, Kim SH (2019) Improved characteristics of MOS interface between http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nanoscale Research Letters Springer Journals

Understanding the Sidewall Passivation Effects in AlGaInP/GaInP Micro-LED

Download PDF
9 pages

Loading next page...
 
/lp/springer-journals/understanding-the-sidewall-passivation-effects-in-algainp-gainp-micro-VfXAICkCzC

References (37)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
ISSN
1931-7573
eISSN
1556-276X
DOI
10.1186/s11671-022-03669-5
Publisher site
See Article on Publisher Site

Abstract

The passivation effects of sulfur treatment and Al O passivation for AlGaInP/GaInP red micro‑light ‑ emitting‑ diodes 2 3 (LEDs) were investigated in terms of the external quantum efficiency (EQE) and the current density showing the peak EQE (J ). We systematically compared the electrical and optical characteristics of the micro‑LEDs with and with‑ EQE, peak out passivation according to various sizes. Interestingly, our investigation indicated that simple electrical characteris‑ tics such as current density–voltage property are difficult to precisely reflect the minor change in electrical properties due to passivation when the device has the inherently low leakage current. Whereas the EQE was enhanced by 20% and J was largely shifted to a lower current density region at the LED with a size of 15 × 15 μm . To examine the EQE, peak passivation effects, we carefully analyzed the EQE and J with the ABC recombination model, and established EQE, peak the methodology to investigate the influence of a sidewall in micro ‑LEDs. As a result, we extracted the surface recom‑ bination velocity regarding the surface passivation, showing a nearly 14% reduction with the passivation. Keywords: Micro‑LED, Passivation, AlGaInP/GaInP MQW etching (ICP-RIE), play a critical role as a Shockley– Introduction Read–Hall (SRH) recombination center. It results in the Inorganic micro-light-emitting-diodes (LEDs) are in the degradation of the EQE value especially for micro-LEDs spotlight for ultra-high resolution display devices due to because the smaller LEDs have a large surface-to-volume its high performance such as high brightness, scalabil- ratio as the device size decreases. Also it could cause the ity, and contrast ratio [1, 2]. In order to implement the current density showing the maximum EQE (J ) energy-efficient micro-LED display, high external quan - EQE, peak shift to a higher current density region [6, 7]. tum efficiency (EQE) should be achieved with a scaled Therefore, to suppress the sidewall induced perfor - pixel smaller than 50 × 50  μm which is a 1/100 size of mance degradations, the various passivation strate- a conventional LED [3]. Furthermore, a maximum EQE gies have been studied for inorganic micro-LEDs. For should be positioned at the relatively low injection cur- instance, the KOH treatment followed by an atomic layer rent regime, which could lead to a strong benefit of a deposition system (ALD) has been reported to passivate power consumption for micro-display compared with the InGaN micro-LED. With passivation, the light emis- the LED for general lighting [4, 5]. However, the sidewall sion uniformity was improved, the EQE reduction by defects, which are mainly caused by the mesa formation shrinking micro-LED dimensions was effectively reduced process such as inductively-coupled plasma reactive ion [8]. In the case of the AlGaInP/GaInP multi-quantum well (MQW) based red micro-LEDs, it is known to have a larger surface recombination velocity (SRV) than InGaN/ *Correspondence: gdmgdm@kaist.ac.kr; shkim.ee@kaist.ac.kr Dae‑Myeong Geum and Sanghyeon Kim are co ‑ corresponding authors GaN MQW, resulting in significant performance degra - and they contributed equally. dation with a pixel size scaling [9]. However, there are School of Electrical Engineering, Korea Advanced Institute of Science still few researches on the AlGaInP/GaInP passivation, and Technology (KAIST ), Daejeon 34141, Republic of Korea Infromation and Electronics Research Institute, Korea Advanced Institute although red LED is one of the important building blocks of Science and Technology (KAIST ), Daejeon 34141, Republic of Korea © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. Park et al. Nanoscale Research Letters (2022) 17:29 Page 2 of 9 for fabricating future full-color micro-displays [10, 11]. for 10  min to reduce the contact resistance of ohmic Furthermore, most of the reported micro-LEDs showed contacts. a high leakage current, indicating the fabrication process After the standard device fabrication, we carried out has not much been optimized than InGaN/GaN LED. the surface passivation. The micro-LEDs were dipped On the other hand, the investigation of EQE relative to in HCl: DI (1:5) solution for 30  s to remove the native injection current density has not been conducted yet in oxide of the sidewall. Then, the sample was dipped in AlGaInP/GaInP MQW micro-LEDs because a low cur- (NH ) S :DI (1:10) solution for 10  min and was subse- 4 2 x rent injection regime would not be critically considered quently loaded into the atomic layer deposition (ALD) for lighting applications. Therefore, to fabricate the highly chamber to minimize the naturally formed native oxide. efficient micro-displays, the exploration of the fabrication Finally, after depositing the 10  nm Al O passivation 2 3 technology of AlGaInP/GaInP LED including pixel for- layer, the passivated micro-LEDs were analyzed com- mation process as well as surface passivation should be paratively with non-passivated micro-LED. The elec - carried out. In addition, the investigation of EQE relative trical characteristics were measured by Keithley 4200. to injection current density including J should be For measuring optical characteristics, we firstly used EQE, peak conducted to develop the red-pixels of display applica- the integrated sphere system. However, when using the tion and understand the effect of sidewall degradation. integrated sphere, inevitable photon loss occurred since In the present work, we fabricated the AlGaInP/ the micro-LED is located a few inches away from the GaInP red micro-LEDs and systematically investigated inner surface of the sphere. Therefore, to measure the the EQE with various device sizes from 15 × 15  μm to lower current density region, we used the photodetec- 80 × 80  μm . Furthermore, to investigate the passivation tor located directly above the micro-LED and calibrated effect for AlGaInP/GaInP micro-LEDs, we conducted the the measured photocurrent to light output power (W) by conventional surface passivation with sulfur treatment referring to the integrated sphere data (Additional file  1: followed by Al O deposition [12–14]. Furthermore, the Figure S1). 2 3 effect of the passivation was carefully examined in terms of the EQE enhancement and J shift through opti- Results and Discussion EQE, peak cal and electrical characterizations. Then, to quantitively Electrical and Optical Characterization of the Fabricated extract the SRV values, by fitting the measured EQE data Micro‑LEDs with the ABC model of recombinations, the SRH recom- The electrical characteristics of the AlGaInP/GaInP bination constants were well analyzed with respect to the micro-LED without passivation were shown in Fig.  1b micro-LED sizes. depending on the device sizes from 15 × 15  μm to 80 × 80  μm . At the reverse bias region, the dark current Methods densities were measured nearby the measurement floor, Fabrication Process of Micro‑LEDs while it was clearly noted that there is a low leakage cur- The epitaxial layers were grown by metal–organic chemi - rent that originates from parasitic current paths such as cal vapor deposition (MOCVD) on a 4-inch semi-insu- a sidewall [15]. Comparing the 15 × 15  μm device char- lating GaAs substrate. The micro-LED structure based acteristics with the same sized device of Oh et al. our IV on AlGaInP/GaInP MQW structure is shown in Fig.  1a. plot shows low leakage current and dark current density The micro-LED fabrication process was started with the level. It suggested that there was negligible degradation standard cleaning process with acetone, methanol, and through various fabrication processes. Also, the system- deionized (DI) water. Sequentially, the mesa isolation atic current density increase over 2 V bias is found as the was conducted to define the pixel by ICP-RIE. We etched device size decreases. In Fig.  1c, the J–L (Current den- the AlGaInP/GaInP epilayers with C l and Ar gas flow sity–Light output power) characteristics of 20 × 20  μm in an ICP-RIE system. In this step, the etching was car- are shown by comparing the same size device of Wong ried out until the middle of the n-GaInP layer. Then, the et  al. [10]. Compared to the device of Wong et  al. the samples were etched to completely remove the remain- 20  μm micro-LED of this paper showed 2.32  μW, which + 2 ing n-GaInP layers and expose the n -GaAs contact is 46 times higher at 20 A/cm . And 6.0 μW which is 54 layer using the H PO : HCl (3:1) solution. For preventing times at 40 A/cm . They passivated the devices by depos - 3 4 electrical short problem of p-contact metal, Al O layer iting Al O followed by nitrogen plasma surface treat- 2 3 2 3 was patially defined for p-conact region. After that, Cr/ ment. Even comparing with the passivated micro-LED, Au (25/75  nm) was deposited for p-type contact metal our micro-LED showed 8 times higher at 20  A/cm and to GaP by the electron beam (e-beam) evaporator. Also, 8 times higher at 40  A/cm . As a result, our micro-LED Pd/Ge/Au (20/40/100  nm) was deposited on the n - device showed a much larger light emission. The inset GaAs layer. Finally, the sample was annealed at 200  °C of Fig.  1d shows the microscope image with 100  A/cm P ark et al. Nanoscale Research Letters (2022) 17:29 Page 3 of 9 Fig. 1 a The process flow of AlGaInP/GaInP MQW LED with passivation. b The current density–voltage (J–V) characteristic of AlGaInP/GaInP MQW LED depending on device sizes. c The current density–light output power (J–L) characteristics of 20 × 20 μm micro‑LED comparing with the same size devices of Ref [10] d Electroluminescence of AlGaInP/GaInP MQW LED with different current density. The inset in (d) is the microscope image of 2 2 20 × 20 μm AlGaInP/GaInP MQW LED at 100 A/cm current density injected to micro-LEDs. When the cur- measurements were conducted for non-passivated and 2 2 rent of 100 A/cm is driven, the red light emission to the passivated micro-LEDs with 15 × 15 μm . In Fig. 2a, there top surface of the 20 × 20  μm device is clearly shown. were almost the same current density–voltage curves at As shown in Fig.  1d, the electroluminescence (EL) spec- the whole bias range. For other device sizes, it was noted tra of 15 × 15  μm LEDs are shown with the peak wave- that there were only negligible changes (not shown here). length 628  nm with current densities in ranges from 53 Compared with other reports having relatively high leak- to 400  A/cm . Resulting full-width at half maximum age current, despite passivation, the reduction of leak- (FWHM) value at 100  A/cm is 14  nm, which is simi- age current was not noticeable in our micro-LEDs [7, lar to previous reported LEDs [16]. Furthermore, it was 10, 11]. Specifically, in Figs.  1b and 2a in our study, the −9 found that the light intensity increases with an increase leakage current of reverse bias was saturated from 10 −7 2 of the injection current density, and without the severe to 10  A/cm in all device dimensions. The leakage cur - peak wavelength change. From these results, the fabri- rent showed a very low level and the saturation behavior cated micro-LEDs are well-functioned with our fabrica- was very stable with increasing reverse bias. Additionally, tion process. in Figs. 1b and 2a, the J–V characteristics were shown in To examine the passivation effects on the electri - the log scale. Therefore, we can check there is no addi - cal characteristics, current density–voltage (J–V) tional leakage current path nearby the threshold voltage Park et al. Nanoscale Research Letters (2022) 17:29 Page 4 of 9 Fig. 2 a J–V curves of non‑passivated and passivated 15 × 15 μm LED. b Diode ideality factor depending on device sizes. c The EQE comparison of 2 2 AlGaInP/GaInP LEDs which have the size of 15 × 15 μm and 80 × 80 μm with and without passivation. d The enhancement ratio of maximum EQE with passivation depending on the device sizes −1 at forward bias. These results strongly suggest that our q ∂ ln I n = (1) devices inherently showed low leakage current level at kT ∂V both reverse and forward bias, thereby, the negligible variation for electrical properties was observed even after where n, q, k, T, I, and V are the ideality factor, the ele- the surface passivation rather than other reports. It could mentary charge, the Boltzman constant, the absolute be attributed to the inherently low leakage current with temperature, current, and the applied voltage, respec- well-defined fabrication process, which could lead to very tively. Figure  2b illustrates the ideality factors of non- small deviation of leakage currents. passivated and passivated micro-LEDs depending on 2 2 For the in-depth analysis, the diode ideality factor was the device sizes from 15 × 15  μm to 100 × 100  μm . The calculated by Eq. (1) value of the 425 × 425  μm LED without passivation was also added for a large-size LED. Referring to the conven- tional interpretation of ideality factor in micro-LEDs, the P ark et al. Nanoscale Research Letters (2022) 17:29 Page 5 of 9 ideality factor near unity means a radiative recombina- Figure 2c shows the normalized EQEs with and without 2 2 tion dominance, while the ideality factor of 2 indicates passivation for 15 × 15 μm and 80 × 80 μm micro-LEDs. SRH recombination dominance through trap states [17]. The values of EQE are normalized by the maximum val - As depicted in Fig.  2b, the value of the ideality factor of ues of passivated micro-LEDs for each size. The enhance - the non-passivated device is from 1.82 (425 × 425 μm ) to ment of the peak EQE values was clearly observed in the 1.93 (80 × 80 μm ). The severe degradation of ideality fac - optical measurement unlike the electrical characteristics. tors depending on mesa sizes is not found compared to Even though the electrical characteristics didn’t show previous reports. Additionally, ideality factors of the pas- definite difference because of the low leakage current sivated micro-LEDs are from 1.89 (80 × 80  μm ) to 1.94 level, the optical performance showed clear improvement (15 × 15  μm ), which are very similar to non-passivated by utilizing the passivation process. LEDs. When compared to other reports, the notice- To compare the effects of surface passivation depend - able increase of ideality factors was not observed with a ing on a different size, the enhancement ratio is defined reduction of micro-LED sizes and passivation. Although by Eq. (2). Max. EQE(passivated) − Max. EQE(non passivated) (2) Enhancement ratio = Max. EQE(non passivated) The surface passivation increased the maximum EQE the ideality factor has been used as a good indirect indi- of 15 × 15  μm micro-LED as 19.8% and the maximum cation for electrical characteristics of LED, it was not EQE of 80 × 80  μm as a 2.4%. Because of the higher sur- so effective to analyze the fabricated LEDs in this work. face-to-volume ratio, 15 × 15  μm was more affected by The difference of ideality factors depending on the device surface recombination of sidewall defects, so the passiva- sizes are subtle like within 2 decimal places, thus it was tion effect was larger than 80 × 80  μm . Besides, calcu- quite difficult to evaluate the sidewall degradation or pas - lated enhancement ratios for the different sizes is shown sivation effects from these values. Because our micro- in Fig. 2d. The 60 μm and 80 μm micro-LEDs showed an LED devices showed an inherently low leakage current, enhancement ratio of less than 5%, but the device pitch which is close to measurement limits, the minor change smaller than 40  μm had a higher value over 10%. The in electrical characteristics with or without passivation enhancement ratio starts to increase rapidly under the couldn’t be measured. device size of 40 μm. The large enhancement of EQEs in In the fabrication process, the etching process using the smaller size micro-LEDs strongly suggests that our BOE and HCl was included. The buffered oxide etchant passivation could suppress the surface recombination (BOE) and HCl have been reported as a good remover for the AlGaInP/GaInP micro-LEDs smaller than 40  μm of the native oxide at the surface of the semiconduc- size. On the other hand, the EQE enhancement might be tor. Thus, the electrical characteristics may look better attributed to the increased LEE. Due to the lower refrac- immediately after immersion in the solutions. However, tive index of Al O, Al O deposition decreases the dif- 2 3 2 3 if there is no additional treatment like sulfur immer- ference of refractive index between LED epilayers and air. sion, the sidewall will be gradually re-oxidized after Therefore, less amount of photons would be guided in the enough time passes [18]. Because the micro-LEDs with- epi-layers due to less total internal reflection. The LEE out passivation in the manuscript are measured after enhancement effect and the passivation effect seem to be being exposed to air quite a few hours later, it doesn’t mixed in the results. To clarify this issue, we simulated seem the wet treatments critically change the electrical 40  μm micro-red-LED (GaP/Al(Ga)InP/MQW/Al(Ga) measurement. Especially, HCl has been widely utilized InP/GaAs substrate) by finite-difference time-domain as an etchant for AlGaInP epilayers or GaInP etch stop (FDTD) simulation of LUMERICAL. The dipole source layer combining with H PO when fabricating the red- 3 4 with the wavelength 625  nm is positioned in the center LEDs [19–21]. However, those studies have not shown of MQWs and the transmittance towards the top and special electrical characteristics or high external quan- sidewall is calculated when 10  nm Al O is deposited or 2 3 tum efficiency like this paper. This indirectly indicates not. As a result, the device with Al O had a 1% decrease 2 3 that the usage of the chemicals such as H PO , HCl, 3 4 in top transmittance and no difference at the sidewall, BOE is not the main factor influencing the micro- which means if the thickness is not optimized, there is LED performance. Also, all wet processes were equally no LEE enhancement. Therefore, the EQE improvements addressed to each sample, thus, the impact of the passi- of this work are not because of enhanced LEE by Al O 2 3 vation steps was fairly compared and the improvement by passivation steps is reasonable. Park et al. Nanoscale Research Letters (2022) 17:29 Page 6 of 9 Fig. 3 a Size‑ dependent EQE of AlGaInP/GaInP micro‑LED. b Peak current density of LED devices (symbol) and the current density range 1% apart from peak current density (line) with (left) and without passivation (right). c The EQE plot fitting based on the ABC model. d Size ‑ dependent SRH parameters were used in fitting the EQE curves with and without passivation. The surface recombination velocity is extracted from the slope deposition but because of the reduced SRH recombina- The Extraction of a Surface Recombination Velocity tion by passivation process. for Micro‑LEDs Although the EQE enhancements were observed, To quantitatively confirm the sidewall contribution, it is through a simple optical characterization, we could necessary to evaluate the passivation effect not just by the relatively evaluate the effects of the passivation such simple electrical characteristics but by J analysis EQE, peak as enhancement ratios. Thus, to compare the sidewall with optical measurement. J can be a useful ana- EQE, peak contribution of micro-LEDs, size-dependent electrical lytical tool to analyze the impact of SRH recombination. characterization, other than J–V curves or the ideality Because J is proportional to A/C , in which A EQE, peak factor, and optical characterization should be simulta- is the SRH recombination coefficient and C is the auger neously considered. recombination coefficient respectively [22]. In Fig.  3a, the EQE of micro-LED without passivation is shown as a function of the current density with various P ark et al. Nanoscale Research Letters (2022) 17:29 Page 7 of 9 sizes. In the literature, when the sidewall degradation has η Bn inj η = η η , η = , (3) ext lee int int a significant impact on micro-LED, the smaller devices 2 3 An + Bn + Cn have lower EQE because of higher sidewall degradation [7]. However, in Fig.  3a the peak EQE is interestingly 2 3 J = qw An + Bn + Cn increased when the size is varied from 80 to 40 μm, and (4) the peak EQE is decreased with a decrease the size from 40 to 15  μm. These results indicate a relatively lower impact of sidewall degradation than other reports due to A = A + v (5) 0 s the well-designed fabrication process and the impact of increased light extraction efficiency (LEE). Here, η , η , η , η , and n are the external quan- ext lee int inj As the mesa size is decreased, the EQE curve shifts to tum efficiency, the light extraction efficiency (LEE), the a higher current density region, and J becomes internal quantum efficiency, the injection efficiency of EQE, peak higher. The 15 × 15 μm device showed J at 444 A/ the LED, and the carrier concentration, respectively. EQE, peak 2 2 cm , whereas relatively large 80 × 80  μm device showed The ABC model assumes that the A, B, C constants and 27  A/cm . It shows a very significant feature of the deg - η of the LED chip are independent of the carrier con- lee radation associated with the size scaling of the LED, centration and operating current. The η is affected lee indicating that the analysis of J would be a key to by factors that can influence the optical path of emit - EQE, peak capture the impact of the surface state at the sidewall. ted photons such as device structures and layer thick- In Fig. 3b, the bar plot shows the range of current den- ness. Those factors don’t change when the miro-LEDs sity which has the EQE value within 1% deviation at the are operated with different current injection. Therefore, maximum point. The diamond symbol points to J . the LEE can be considered as constant. Also, the η is EQE, peak inj Dry etched sidewall surface could have a non-uniform deserved as 100% at the low current density region in surface condition in terms of defect density (damage). the MQW [23]. Therefore, considering constant η and lee Therefore, before the annihilation of the defective region η , the experimental normalized EQE plot can be fitted inj by the surface passivation, this non-uniformity might by the normalized η in Eq.  (3). The current density of int contribute to the variation of the J , as we meas- the ABC model is calculated by Eq.  (4). J and w are the EQE, peak ured. Therefore, we indicated J value with error current density and the width of the quantum well. For EQE, peak bar. The black and red plots represent the data without the w value, 6 nm was used referring to the thickness of passivation and with passivation for each. Comparing the the quantum well of the epitaxial structure used in this two plots, the J shift was reduced with the passiva- study. The carrier concentrations (n) were swept from EQE, peak 15 19 −3 tion process. J of the 15 × 15  μm device without 10 to 10  cm to calculate EQE as a function of the EQE, peak the surface passivation ranges from 347 to 666  A/cm . current density on a log scale. Based on those equations, With the surface passivation, the range is lowered from the normalized measured EQE was fitted by the calcu - 2 2 302 to 444  A/cm . On the other hand, the 40 × 40  μm lated internal quantum efficiency of the ABC recom - device is relatively less affected than a 15 × 15 μm device bination model. As the most governing parameter, we by sidewall recombination. Furthermore, the 60 × 60 μm varied the values of the A value to investigate the impact and 80 × 80  μm size devices showed a slightly lowered of the surface recombination. On the other hand, we −10 3 −1 peak current density, but the effect was negligibly small. fixed the B and C coefficient as 1.5 × 10  cm  s and −29 6 −1 Therefore, we can conclude that the sidewall contribu - 4 × 10  cm  s since those parameters have minor tion of AlGaInP/GaInP micro-LED is very critical under changes on fitting results of the target current density the size of 40 × 40  μm , which is the dimension of our range (Additional file  2: Figure S1)  [24, 25]. In Fig.  3c, interest for the display to achieve high EQE and maintain the normalized EQE of 20 × 20  μm is fitted by different 7 −1 7 −1 peak current density. Also, it can be effectively reduced A values from 4.2 × 10  s to 6.2 × 10  s . Increasing A through wet treatment followed by depositing the Al O parameter moves to the J to higher current den- 2 3 peak, EQE passivation layer. sity, reflecting the increased SRH recombination at the To deeply investigate the origin, analyzing the effi - sidewall. Therefore, we chose the reasonable A values of ciency behavior by fitting through the ABC model is a good fitting results which induce less than 5% fitting dif - useful strategy. The A, B, and C constant represents the ference between experimental data and calculated result. SRH non-radiative recombination, the radiative recom- The fitting difference is defined as the ratio (%) of the dif - bination, and the Auger recombination for each. In the ference between normalized EQE and normalized IQE ABC model, the external quantum efficiency is expressed to normalized EQE. The best fitting result is plotted by 7 −1 as follows [22]. 5.20 × 10  s of A parameter in Fig.  3c. Our target fit - ting region is the current density around J and EQE, peak, Park et al. Nanoscale Research Letters (2022) 17:29 Page 8 of 9 under J . This region is where SRH recombination amount of the trap along with the energy level will fur- EQE, peak (A parameter) has a dominant impact on IQE calcula- ther improve the performance. tion and the magnitude of sidewall degradation can be shown. At high current region, because there are other Conclusion hindrances such as current crowding effect or high series In conclusion, we fabricated the AlGaInP/GaInP micro- resistance, it is hard to obtain perfectly matched fitting LEDs and investigated the electrical and optical charac- results. Therefore, we fitting process was focused on 2 teristics of the devices with the sizes from 15 × 15  μm the low current density region including J where 2 EQE, peak to 80 × 80  μm . In this process, the specific EQE value LEDs show the inherent performances without additional of the Red LED pixel was investigated to a low current effects. The yellow and blue shadowing region represents density range and this data will be an important refer- which recombination mechanism is dominant. ence aimed at using the AlGaInP/GaInP red LED as a The constant A which is the SRH recombination coef - unit pixel micro-LED display. Also, the investigation on ficient can be re-written with the bulk SRH coefficient the effect of surface passivation with sulfur treatment A the surface recombination velocity (SRV) v , p erim- 0, followed by Al O deposition was conducted. Since our 2 3 eter P and area S in Eq.  (5). The bulk term is A which micro-LEDs showed the inherently low leakage current describes that the SRH recombination at the bulk epi-lay- before passivation with well-arranged fabrication pro- ers, and the perimeter term is v · which describes the cess, there was no notable electrical difference with and SRH recombination at the device surface. To focus on the without passivation. On the other hand, the clear EQE quantitative value of v , we utilized the A, B, C recombi- improvement and lowered J was observed par- EQE, peak nation coefficients as keys to connect the experimental 2 ticularly in micro-LED with a size under 40 × 40  μm . data and SRH recombination coefficient. Therefore we This motivated us to suggest a new analysis method to can extract the SRV by calculating the slope of A param- provide an insight into the surface state of the LEDs. eter versus P/S ratio depending on the device size by Therefore, we first introduced the novel analysis meth - Eq.  (5). In Fig.  3d, the SRH recombination constant (A) odology with J to evaluate the sidewall effect in EQE, peak of with and without passivation are shown with the error micro-LED. Additionally, we fitted the experimental bars. The median values of each range showed the best- EQE with the ABC recombination model and extracted matched fitting results and they were used to calculate the SRV from the size-dependent SRH constant (A), SRV. In the case of passivation, the value of 27,445 cm/s and demonstrated a 14% improvement with the passi- was 15% lower than that of 32,439 cm/s without passiva- vation. The evaluation process shown in this paper can tion by extracting the slopes. Although the extracted SRV be used as a straightforward, but simple guideline to values were not be decreased significantly, it is worth analyze the sidewall-related performance degradation noting that the change of J shift depending on EQE, peak or improvement of micro-LED. Consequently, to real- the various micro-LED size allows us to extract the SRV ize high EQE red micro-LED, further studies that can values. Although the extracted SRV values were not be remove the trap sites at midgap and/or near the valence decreased significantly, it is worth noting that the change band should be carried out in the future. of J shift depending on the various micro-LED EQE, peak size allows us to extract the SRV values. Abbreviations Consequently, even if we achieved the peak EQE Micro‑LED: Micro ‑light ‑ emitting‑ diode; EQE: External quantum efficiency; SRV: enhancement, the lower J , and the SRV reduc- EQE, peak Surface recombination velocity; ICP‑RIE: Inductively‑ coupled plasma reactive tion, the enhancement of device performance was not ion etching; SRH: Shockley–Read–Hall; ALD: Atomic layer deposition; MQW: Multi‑ quantum well; MOCVD: Metal–organic chemical vapor deposition; so dramatic, indicating further study should be care- DI: Deionized; EL: Electroluminescence; FWHM: Full with at half maximum; fully designed and carried out to maximize the perfor- BOE: Buffered oxide etchant; FDTD: Finite ‑ difference time ‑ domain; LEE: Light mance of scaled LEDs. The possible reason for limited extraction efficiency. enhancement in this work is the incomplete removal of the defects at the sidewall. Since the sulfur treatment and Supplementary Information the Al O passivation are known to remove the native The online version contains supplementary material available at https:// doi. 2 3 org/ 10. 1186/ s11671‑ 022‑ 03669‑5. oxide of GaAs and reduce the surface density states near the conduction band edge [28], thereby, the surface Additional file 1: Supplementary information for optical measurement trap states at the midgap and/or near the valence band for low current density: The photodetector measurement. can still be remained and limit the device performance Additional file 2: Supplementary information for selection fitting enhancement even with the passivation. This suggests parameter. that further passivation study with a consideration of the P ark et al. Nanoscale Research Letters (2022) 17:29 Page 9 of 9 Authors’ Contributions In Ga As and insulator by H annealing with Pt gate electrode. Appl 0.53 0.47 2 PJH, BWJ, were involved in experiments. PJH, BWJ, GDM, and SHK participated Phys Lett 115:143502 in analysis and discussions. PJH, GDM, and SHK drafted the manuscript. All 14. Kim S‑H, Yokoyama M, Nakane R, Ichikawa O, Osada T, Hata M, Takenaka authors read and approved the final manuscript. M, Takagi S (2014) High performance tri‑ gate extremely thin‑body InAs‑ On‑Insulator MOSFETs with high short channel effect immunity and Vth Funding tunability. IEEE Trans Electron Devices 61(5):1354–1360 This work was supported by Samsung Research Funding & Incubation Center 15. Cchubert EF (2006) Ligh‑ emitting diodes, 2nd edn. Cambridge University of Samsung Electronics under Project number SRFC‑ TB2003‑02. Press, Cambridge, pp 65–66 16. Zhao Y, Liang J, Zeng Q, Li Y, Li P, Fan K, Sun W, Lv J, Qin Y, Wang Q, Tao J, Availability of Data and Materials Wang W (2021) 2000 PPI silicon‑based AlGaInP red micro ‑LED arrays fabri‑ The datasets used and analysed during the current study are available from cated via wafer bonding and epilayer lift‑ off. Opt Express 29:20217–20228 the corresponding author on reasonable request. 17. Sah CT (1991) Fundamentals of solid‑state electronics. World Scientific, Singapore, pp 438–443 18. Petrovykh DY, Yang MJ, Whitman LJ (2003) Chemical and electronic Declarations properties of sulfur‑passivated InAs surfaces. Surf Sci 523:231–240 19. Tseng MC, Chen CL, Lai NK, Chen SI, Hsu TC, Peng YR, Horng RH (2014) Competing interests P‑side ‑up thin‑film AlGaInP ‑based light ‑ emitting diodes with direct The authors declare that they have no competing interests. ohmic contact of an ITO layer with a GaP window layer. Opt Expres 22:A1862–A1867 Received: 23 December 2021 Accepted: 15 February 2022 20. Han S, Xu C, Li H, Liu S, Xu H, Zhu Y, Fang A, Wang X (2021) AlGaInP‑based Micro‑LED array with enhanced optoelectrical properties. Opt Mater 114:110860 21. Chilukuri K, Mori MJ, Dohrman CL, Fitzgerald EA (2006) Monolithic CMOS‑ compatible AlGaInP visible LED arrays on silicon on lattice‑ engineered References substrates (SOLES). Semicond Sci Technol 22(2):29–34 1. Kim SH, Kim SK, Shim JP, Geum DM, Ju GW, Kim HS, Lim HJ, Lim HR, Han 22. Smith JM, Ley R, Wong MS, Baek YH, Kang JH, Kim CH, Gordon MJ, Naka‑ JH, Lee SB, Kim HS, Bidenko P, Kang CM, Lee DS, Song JD, Choi WJ, Kim HJ mura S, Speck JS, Denbaars SP (2020) Comparison of size‑ dependent (2018) Heterogeneous integration toward a monolithic 3‑D chip enabled characteristics of blue and green InGaN microLEDs down to 1um in by III–V and Ge materials. IEEE J Electron Devices Soc 6:579–587 diameter. Appl Phys Lett 116:071102 2. Geum DM, Kim SK, Kang CM, Moon SH, Kyhm JH, Han JH, Lee DS, Kim SH 23. Zhao HP, Liu GG, Zhang J, Arif RA, Tansu N (2013) Analysis of internal (2019) Strategy toward the fabrication of ultrahigh‑resolution micro ‑LED quantum efficiency and current injection efficiency in III‑nitride light ‑ displays by bonding‑interface ‑ engineered vertical stacking and surface emitting diodes. J Display Technol 9:212–225 passivation. Nanoscale 11:23139–23148 24. Levinshtein M, Rumyantsev S, Shur M (1999) Handbook series on semi‑ 3. Wu T, Sher C‑ W, Lin Y, Lee C‑F, Liang SJ, Lu YJ, Chen S‑ WH, Guo WJ, Kuo conductor parameters. Tenary and quaternary III–V compounds, vol 2. H‑ C, Chen Z (2018) Mini‑LED and micro ‑LED: promising candidates for World Scientific, Singapore the next generation display technology. Appl Sci 8(9):1557 25. Brendel M, Kruse A, Jönen H, Hoffmann L, Bremers H, Rossow U, 4. Day J, Li J, Lie DYC, Bradford C, Lin JY, Jiang HX (2011) III‑Nitride full‑scale Hangleiter A (2011) Auger recombination in GaInN/GaN quantum well high‑resolution microdisplays. Appl Phys Lett 99:031116 laser structures. Appl Phys Lett 99:031106 5. Chong WC, Cho WK, Liu ZJ, Wang CH, Lau KM (2014) 1700 Pixels per 26. Olivier F, Daami A, Licitra C, Templier F (2017) Shockley–Read–Hall and inch (PPI) passive‑matrix micro ‑LED Display powered by ASIC. In: IEEE Auger non‑radiative recombination in GaN based LEDs: a size effect compound semiconductor integrated circuit symposium (CSICS). pp 1–4 study. Appl Phys Lett 111:022104 6. Tian P, McKendry JJD, Gong Z, Guilhabert B, Watson IM, Gu E, Chen Z, 27. Verzellesi G, Saguatti D, Meneghini M, Bertazzi F, Goano M, Meneghesso Zhang G, Dawson MD (2012) Size‑ dependent efficiency and efficiency G, Zanoni E (2013) Efficiency droop in InGaN/GaN blue light ‑ emitting droop of blue InGaN micro‑light emitting diodes. Appl Phys Lett diodes: physical mechanisms and remedies. Appl Phys Lett 114:071101 101:231110 28. Zhou L, Bo B, Yan XZ, Wang C, Chi YD, Yang XT (2018) Brief review of 7. Oh JT, Lee SY, Moon Y T, Moon JH, Park SW, Hong KY, Song KY, Oh CH, surface passivation on III–V semiconductor. Curr Comput‑Aided Drug Des Shim JI, Jeong HH, Song JO, Amano H, Seong TY (2018) Light output 8(5):226 performance of red AlGaInP‑based light emitting diodes with different chip geometries and structures. Opt Express 26:11194–11200 Publisher’s Note 8. Wong MS, Lee CM, Myers DJ, Hwang D, Kearns JA, Li T, Speck JS, Naka‑ Springer Nature remains neutral with regard to jurisdictional claims in pub‑ mura S, Denbaars SP (2019) Size‑independent peak efficiency of III‑nitride lished maps and institutional affiliations. micro‑light ‑ emitting‑ diodes using chemical treatment and sidewall passivation. Appl Phys Express 12:097004 9. Bulashevich KA, Karpo SY (2016) Impact of surface recombination on efficiency of III‑nitride light ‑ emitting diodes. Phys Status Solidi RRL 10(6):480–484 10. Wong MS, Kearns JA, Lee CM, Smith JM, Lynsky C, Lheureux G, Choi HS, Kim JW, Hon Kim C, Nakamura S, Speck JS, DenBaars SP (2020) Improved performance of AlGaInP red micro‑light ‑ emitting diodes with sidewall treatments. Opt Express 28:5787–5793 11. Huang HH, Huang SK, Tsai YL, Wang SW, Lee YY, Weng SY, Kuo HC, Lin CC (2020) Investigation on reliability of red micro‑light emitting diodes with atomic layer deposition passivation layers. Opt Express 28(25):38184–38195 12. Suzuki R, Taoka N, Yokoyama M, Lee S, Kim SH, Hoshii T, Yasuda T, Jevasu‑ wan W, Maeda T, Ichikawa O, Fukuhara N, Hata M, Takenaka M, Takagi S (2012) 1‑nm‑ capacitance‑ equivalent‑thickness HfO /Al O /InGaAs metal‑ 2 2 3 oxide‑semiconductor structure with low interface trap density and low gate leakage current density. Appl Phys Lett 100:132906 13. Kim SK, Geum D‑M, Lim H‑R, Kim HS, Han J‑H, Hwang DK, Song JD, Kim H‑ J, Kim SH (2019) Improved characteristics of MOS interface between

Journal

Nanoscale Research Letters – Springer Journals

Published: Mar 1, 2022

Keywords: Micro-LED; Passivation; AlGaInP/GaInP MQW

There are no references for this article.