Hexagonal BN‐Assisted Epitaxy of Strain Released GaN Films for True Green Light‐Emitting Diodes

Abstract Epitaxial growth of III‐nitrides on 2D materials enables the realization of flexible optoelectronic devices for next‐generation wearable applications. Unfortunately, it is difficult to obtain high‐quality III‐nitride epilayers on 2D materials such as hexagonal BN (h‐BN) due to different atom hybridizations. Here, the epitaxy of single‐crystalline GaN films on the chemically activated h‐BN/Al2O3 substrates is reported, paying attention to interface atomic configuration. It is found that chemical‐activated h‐BN provides B—O—N and N—O bonds, where the latter ones act as effective artificial dangling bonds for following GaN nucleation, leading to Ga‐polar GaN films with a flat surface. The h‐BN is also found to be effective in modifying the compressive strain in GaN film and thus improves indium incorporation during the growth of InGaN quantum wells, resulting in the achievement of pure green light‐emitting diodes. This work provides an effective way for III‐nitrides epitaxy on h‐BN and a possible route to overcome the epitaxial bottleneck of high indium content III‐nitride light‐emitting devices.


DOI: 10.1002/advs.202000917
(LEDs), such as green LEDs. [9,10] On the one hand, the lattice mismatch causes larger residual stress in the multiple quantum wells (MQWs) region, which limits the incorporation of indium (In) in the MQWs. [11,12] On the other hand, the larger residual stress introduces bigger polarization electric field in the MQWs region, resulting in a stronger quantum confinement Stark effect (QCSE), which significantly reduces the luminous efficiency of the green LED. [13] Therefore, the epitaxy of nitrides on 2D materials is considered to be one of the potential methods for preparing highefficient green LEDs. Besides, epitaxy of IIInitrides on 2D materials also shows its advantage for mechanically transferring epitaxial structures onto foreign substrates, thereby obtaining flexible III-nitride based devices at a low cost and vertical devices as well. [14,15] Notably, as a family member of III-nitrides, hexagonal BN (h-BN) owes better growth compatibility than other 2D materials for the epitaxy of III-nitride films and is thus believed to be the most suitable 2D material for III-nitrides epitaxy. [16,17] However, the integration between h-BN and conventional IIInitride film still remains a challenge due to the difficulty in combining different atomic hybridization. [18][19][20][21] In fact, it is very difficult to form covalent bonds on 2D materials due to the absence of dangling bonds. [22][23][24][25] Therefore, nucleation of GaN on 2D materials is almost impossible except that the 2D materials are not perfect. [26,27] A couple of research groups tried to make dangling bonds or defects on 2D materials by using oxygen-plasma pretreatment on the 2D materials to enhance the nucleation of GaN. [28,29] For example, Chung et al. reported the achievement of high-quality GaN films on oxygen-plasmatreated graphene/Al 2 O 3 substrates. [1] Besides, Wu et al. reported the growth of AlGaN-based deep-ultraviolet LED structure on oxygen-plasma-treated h-BN/Al 2 O 3 . [30] Although several groups have reported the epitaxy of III-nitrides on h-BN, graphene, and so on, the interface bonding behaviors and/or the nucleation phenomena of III-nitrides are seldom reported and are not well studied to the best of our knowledge. [31][32][33] In this work, we study the epitaxy of GaN on 2-in. h-BN/Al 2 O 3 , and focus on the interface bonding and nucleation behavior. A chemical activation method is specially used to generate N-O bonds to facilitate the nucleation of GaN on the h-BN surface and thus improve the crystal perfection for epilayers. It is found that the artificially added N-O bonds not only create more sites for following GaN nucleation but also modify the lattice polarity of GaN to be uniform Ga-polarity one, leading to high-quality material with flat surface. Based on those smooth GaN, we are then able to grow high-quality InGaN quantum wells structure for green LEDs. That h-BN layer is found to relax the compressive strain in GaN film and thus improve indium incorporation during the growth of InGaN quantum wells, leading to the emission wavelength redshift and achievement of pure green lightemitting diodes.
GaN epitaxy was performed on 2-in. h-BN grown on (0001) Al 2 O 3 substrate. The typical thickness for the h-BN is about 3 nm. As shown in Figure 1a, a typical Raman scattering spectrum of the 3-nm-thick h-BN shows an intense scattering peak at 1364 cm −1 with a full width at half-maximum (FWHM) as small as 10.7 cm −1 , indicating a sp 2 -hybridized BN with good crystal quality. The h-BN shows wrinkle-like morphology with a root mean square value of 0.9 nm in a scanned area of 5 × 5 µm 2 , as depicted in Figure 1b.
Conventional two-step growth was performed to deposit GaN films on h-BN/Al 2 O 3 , which had achieved great success in the epitaxy of III-nitrides on sp 3 -bonded single crystal substrates such as Al 2 O 3 . [34,35] Unfortunately, with a growth interruption after low temperature (LT)-GaN annealing, only limited number of GaN nucleation islands next to wrinkles has been observed, as depicted in Figure 1c. Afterward, those islands cannot coalesce well during conventional high temperature growth process and the high temperature (HT)-GaN film shows a discontinuous surface, as depicted in Figure 1d. It reveals that direct growth of GaN on h-BN/Al 2 O 3 is difficult, because the absence of dangling bonds in 2D h-BN indeed hinders GaN nucleation, especially, the more perfect the 2D, the worse the nucleation of GaN.
Here, we propose a chemical activation method to generate dangling bonds on the surface of h-BN and thus to improve the nucleation ability of sp 3 -hybrided GaN. To activate the surface, h-BN was treated by hydrochloric acid (HCl) and deionized water, which is expected to introduce the hydroxide (OH) and thus the oxygen bonds so that the nucleation sites can be created. As depicted in Figure 1e, more GaN crystalline grains are observed on the activated h-BN surface. This is further proved by the grain density statistics (see Figure S2, Supporting Information). After HCl treatment, the GaN grain density increases by about threefold from 24 to 69 µm −2 , indicating a significant increase of nucleation sites, i.e., more surface dangling bonds are generated on h-BN surface. This facilitates the coalescence of HT-GaN and further leads to the formation of continuous and smooth films, as shown in Figure 1f.
To reveal the surface-activated mechanism of h-BN after HCl treatment, X-ray photoemission spectroscopy (XPS) measurements were carried to identify the surface chemical states on h-BN. Figure 2a,b presents that the chemical state of B 1s remains B-N (190.8 eV) and the B-O related chemical state (191.4 eV) appears after the HCl treatment. [36,37] Simultaneously, the chemical state of N 1s remains B-N (396.5 eV) and the N-O related chemical state (400.4 eV) are detected after the HCl treatment, as described in Figure 2c,d. [38,39] Meanwhile, XPS spectra of activated h-BN show that the intensities of B 1s and N 1s peaks obviously decrease, which are mostly attributed to the degradation of lattice structure (see Figure S3, Supporting Information). These results indicate that the newly formed surface defects and the original defects will provide some oxygen-related artificial dangling bonds on the h-BN surface during HCl treatment.
Then, we performed first-principles density functional theory (DFT) calculation to confirm the chemical adsorption process of O atoms on the h-BN surface. As shown in Figure 3a,b, four op-       This kind of atomic bond configuration at the interface leads to the epitaxy of Ga-polarity GaN, which is experimentally confirmed after the growth by chemical etching effect (see Figure S4, Supporting Information). This also indicates that the proposed atomic configurations at the interface are reasonable.
The improvement of GaN nucleation on h-BN not only enhances the surface smoothness of GaN epilayers but also improves their crystalline quality, as shown in Figure 5a,b. The FWHMs of X-ray diffraction (XRD) -rocking curve for the symmetric (0002) and asymmetric (1012) planes of 3-µm-thick GaN epilayer decrease from 591 and 841 arcsec to 316 and 543 arcsec, respectively. In fact, the GaN film grown on the activated 3nm-thick h-BN interlayer exhibits similar crystal quality as those grown on Al 2 O 3 at the same growth condition (see Figure S5a, Supporting Information). But there is an advantage when using h-BN interlayer, which is strain relaxation in the GaN layer. Figure 5c,d shows the Raman scattering spectra of GaN growth with and without activated h-BN interlayer. It is shown that strainsensitive E 2 (high) and A 1 (LO) peaks shift from 570.5 to 569.1 cm −1 and from 737.5 to 735.8 cm −1 , respectively, indicating the relaxation of residual compressive strain in the GaN epilayer. [8,47] In fact, the residual compressive strain in the GaN layer can be further relaxed by increasing the thickness of h-BN interlayer (see Figure S5b, Supporting Information). However, the crystalline quality is also degraded with increasing the h-BN thickness unfortunately. Anyway, this strain relaxation in the GaN epilayer is   definitely helpful to improve the emission of following InGaN quantum wells toward longer wavelength.
Finally, LEDs based on InGaN/GaN MQWs were grown on asfabricated undoped GaN/h-BN/Al 2 O 3 templates, where the thickness of activated h-BN and undoped GaN were 3 nm and 3 µm, respectively. For comparison, the LEDs on Al 2 O 3 were grown in one run. Figure 6a,b exhibits XRD 2 -scans for those LED structures, where intense diffraction peaks from GaN epilayer and satellite peaks from the InGaN/GaN MQWs up to the fourth order are observed in both cases, indicating excellent quality and sharp interfaces of the InGaN/GaN MQWs. It is noted that the fitted In composition in the InGaN quantum wells increases from 25% to 28% after using the activated h-BN interlayer. The AlGaN diffraction peaks come from the electron blocking layer (EBL) in the LED structure. To evaluate the relationship of In composition and strain distribution in InGaN/GaN MQWs, the X-ray reciprocal space mapping (RSM) scans of the (1015) asymmetric reflection are measured, as depicted in Figure 6c,d. The satellite peaks from MQWs vertically line up with the GaN diffraction maximum along Q x , demonstrating that the InGaN/GaN MQWs are coherently grown on the bottom n-type GaN layer. As shown by the peak position relative to Q x , the n-GaN layer grown on the h-BN template has a larger in-plane lattice constant than that grown on Al 2 O 3 , indicating a smaller in-plane compressive strain in the n-GaN layer and InGaN/GaN MQWs region. This result indicates that the biaxial compressive strain within the basal plane in GaN is partially released by using h-BN interlayer, which could lead to high incorporation efficiency of indium during the growth of In-GaN/GaN MQWs. Figure 6e,f demonstrates electroluminescence (EL) spectra for the LEDs with and without using the activated h-BN interlayer. The LED grown on the 3-nm-thick h-BN interlayer shows a pure green emission with a center wavelength at ∼ 556 nm at typical injection current density of 40 mA·mm −2 , which is ∼ 26 nm redshift in comparison with that on Al 2 O 3 . It is known that the relaxed compressive strain suppresses the QCSE in InGaN/GaN MQWs, leading to the slight blueshift of emission wavelength and stronger electron-hole wave function overlaps and improved radiative recombination rate. [48,49] According to the estimation of the degree of compressive stress relaxation of green LED, the redshift for ∼ 1.1 cm −1 of the E 2 (high) peak belonging to GaN film below MQWs region will lead to the blueshift of emission wavelength of about 2 nm. [50] However, the blueshift is compensated since strain relaxation is beneficial to the In incorporation in the InGaN wells and it leads to an emission peak redshift of ∼26 nm. The slight broadening of the linewidth is likely caused by In alloy fluctuations in In-richer InGaN wells. By the way, a slight blueshift of emission and spectrum broadening are observed in both LEDs with increasing the injection current density, which is most likely induced by the combined effect of the free-carrier screening of piezoelectric fields and the band filling effect. We also investigated temperature-dependent photoluminescence (PL) spectra of both LED structures (see Figure S6, Supporting Information). It shows that the insertion of the h-BN interlayer not only keeps the emission intensity but also tunes the emission toward a longer wavelength, as what we observed in the EL spectra. Notably, the internal quantum efficiency (IQE) www.advancedsciencenews.com www.advancedscience.com of green LED structure on activated h-BN/Al 2 O 3 is ∼23% (∼556 nm) that is close to ∼25% (∼530 nm) for the one on Al 2 O 3 . Normally, with the increase of emission wavelength, InGaN-based LEDs suffer from a systematic drop in efficiency, known as the green gap in III-nitrides. [51,52] Therefore, it is reasonable to believe that the application of h-BN interlayer provides a promising approach to partially break through the green gap in III-nitrides, i.e., the epitaxial bottleneck of high In content nitrides, by commercial MOCVD technique. Epitaxy

Experimental Section
Fabrication of Crystalline h-BN: The 2-in. (0001) Al 2 O 3 substrates were transferred into a plasma-assisted molecular beam epitaxy (PA-MBE) chamber and then thermally cleaned for 30 min at 900°C. Subsequently, the BN was deposited on the entire surface of the Al 2 O 3 at 900°C with a growth rate of 0.2 nm·min −1 . Through high temperature annealing at 1700°C in N 2 ambient for 2 h, the quality of that BN are much improved (see Figure S1, Supporting Information). That crystalline h-BN was then used as a template for GaN epitaxy.
Chemical Activation of Crystalline h-BN: The h-BN/Al 2 O 3 substrates were put in acetone, ethanol, and deionized water solution (30 mL) for 3 min to clean successively. Then, these h-BN/Al 2 O 3 substrates were dipped into hydrochloric acid solution for 10 min for activation treatment. Finally, the treated samples were rinsed by deionized water and dried with flowing nitrogen.
Epitaxial Growth of GaN Films and InGaN-Based LEDs: GaN films were grown on the activated h-BN/Al 2 O 3 templates by an AIXTRON 3 × 2 in. FT MOCVD system. H 2 was used as the carrier gas. Prior to the deposition, the substrates were thermally cleaned in H 2 ambient at 1100°C for 5 min. The growth began with a low temperature GaN (LT-GaN) nucleation layer deposited at 530°C. Then, the growth temperature ramped up to 1050°C for the growth of the high temperature GaN (HT-GaN)  Characterization: XRD measurement was performed by X'Pert3 MRD system using Cu K 1 X-ray source. Surface morphology was measured by atomic force microscopy (AFM) in tapping mode (Bruker Dimension ICON-PT) and scanning electron microscopy (SEM). XPS was performed to quantitatively estimate the chemical states of h-BN before and after HCl treatment. The XPS spectra were measured by an ESCALab 250 Analytical XPS spectrometer with a monochromatic X-ray source (Al K , h = 1486.6 eV). The binding energies of the spectra were referred to that of the C 1s peak at ∼284.8 eV. Strain in GaN was characterized by Raman scattering spectroscopy. The EL spectrum of the GaN-based LEDs was obtained at a driving current of 70 mA by using the home-made acquisition equipment at room temperature including a lock-in amplifier system and photomultiplier tube. The PL spectra of these LED structures were measured by a PL system which includes a He-Cd laser (325 nm, 30 mW) as an excitation source, a Jobin Yvon iHR550 spectrometer, a Syncerity charge coupled device (CCD), and a closed circle helium cryostat.
Simulation: DFT calculations were implemented by the Vienna ab initio Simulation Package (VASP) code. The projector augmented wave (PAW) pseudopotentials were used for electron-ion interaction. The generalized gradient approximation (GGA) was used to the exchangecorrelation functional as proposed by Perdew-Burke-Ernzerhof (PBE). The energy cutoff for plane wave expansion is 520 eV. In the calculation, (5 × 5 × 1) Monkhorst-Pack k-points were used and spin polarization was considered. This monolayer h-BN unit cell model contained a 20 Å vacuum layer along the c-direction. The lattice parameters of the h-BN unit cell were a = b = 2.513 Å, = = 90°, = 120°. More detailed information was provided in calculation part (i.e., Binding energy computation) of the Supporting Information.
Statistical Analysis: The data were obtained by Nano Measurer 1.2 for Windows. In order to ensure the repeatability of the nucleation experiment, different regions were taken to calculate the nucleation density, and sample size (n) for each statistical analysis was 3. Among the GaN grain density on activated h-BN/Al 2 O 3 is about 69 ± 1.6 µm −2 , and the density on untreated h-BN/Al 2 O 3 is about 24 ± 2.1 µm −2 .

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.