High-Quality Crystal Growth and Characteristics of AlGaN-Based Solar-Blind Distributed Bragg Reflectors with a Tri-layer Period Structure

To realize AlGaN-based solar-blind ultraviolet distributed Bragg reflectors (DBRs), a novel tri-layer AlGaN/AlInN/AlInGaN periodical structure that differs from the traditional periodically alternating layers of high- and low-refractive-index materials was proposed and grown on an Al0.5Ga0.5N template via metal-organic chemical vapour deposition. Because of the intentional design of the AlInGaN strain transition layer, a state-of-the-art DBR structure with atomic-level-flatness interfaces was achieved using an AlGaN template. The fabricated DBR exhibits a peak reflectivity of 86% at the centre wavelength of 274 nm and a stopband with a full-width at half-maximum of 16 nm.

instead of AlGaN 17 . Although the centre wavelength approaches the solar-blind region, the peak reflectivity of 60% is still low because of the difficulty of achieving high-quality BAlN ternary alloys. Zhang et al. very recently reported a solar-blind AlInN/AlGaN DBR grown by molecular beam epitaxy (MBE); however, the stopband of the DBR exhibited a pyramid shape that resulted mainly from the blurry and rough interface between the AlInN and AlGaN layers 18 . The clear and flat interfaces between DBR constitutive layers are known to be critical for highly reflective DBRs. However, because of the low surface migration mobility of Al atoms during growth, MBE conducted at a relatively low growth temperature is not suitable for the growth of high-Al-composition AlGaN DBR multilayer structures with a clear interface; additionally, MBE is much less efficient than the metal-organic chemical vapour deposition (MOCVD) method in fabricating multi-period thick films. In summation, a number of challenges still remain for achieving high-quality solar-blind UV DBRs.
Here, we used a periodic AlGaN/AlInN/AlInGaN tri-layer structure instead of the traditional periodically alternating layers of high-and low-refractive-index materials to fabricate a solar-blind UV DBR. We used a composition-graded AlInGaN transition layer between the AlGaN and AlInN layers to accommodate the lattice mismatch between AlInN and AlGaN. The AlGaN alloy was used as the high-refractive-index layer in the DBR. However, the design of the AlGaN layer must balance competing trade-offs. On the one hand, higher refractive-index contrast will be obtained with a lower Al content in the AlGaN layer, resulting in a high-reflectivity DBR. On the other hand, the low-Al-content AlGaN alloys will increase residual absorption in the stopband, which degrades DBR performance. To retain the advantages of the high-refractive-index contrast and the low residual absorption, a relatively low Al content was chosen under the precondition that the bandgap of AlGaN was larger than the short-wavelength edge of the stopband.

Results
The design of the tri-layer-structured solar-blind UV DBRs is a compromise solution that considers all of the factors of lattice matching, refractive-index contrast, residual absorption, and efficient fabrication. According to these compromise considerations, we designed a 20-period AlGaN/AlInN/AlInGaN tri-layer DBR structure with a calculated centre wavelength of 275 nm. The compositions of the three DBR layers were Al 0.5 Ga 0.5 N, Al 0.96 In 0.04 N, and Al x In 0.02 Ga 0.98−x N (0.75 ≤ x ≤ 0.95), and the optimized thickness of the AlGaN/AlInN/AlInGaN layers calculated using the transfer matrix method was approximately 16/12/28 nm. The refractive index of Al 0.85 In 0.02 Ga 0.13 N was taken as the average value of the Al x In 0.02 Ga 0.98−x N layer for calculating the reflectivity spectra. All of the refractive indices were taken from the literature [19][20][21] .
To reduce the strain resulting from the lattice mismatch, a fully strain-relaxed Al 0.5 Ga 0.5 N template on a sapphire substrate was employed for fabrication of the DBR structure by MOCVD 22 . The cross-sectional transmission electron microscopy (TEM) image of the DBR in Fig. 1 shows that the interfaces between the DBR layers were well-defined and perfectly flat. The thickness of each layer was accurately measured owing to the clear and abrupt interfaces and was found to be 16.0, 10.9, and 27.6 nm for the AlGaN, AlInN, and AlInGaN layers, respectively, as shown in the magnified image presented in the inset of Fig. 1; these thicknesses are consistent with the designed structure. The thickness of the Al 0.5 Ga 0.5 N template measured by TEM was 990 nm; this thickness is not displayed integrally in Fig. 1. Furthermore, the high-resolution TEM image of the DBR in Fig. 2 shows that the high-quality AlGaN/AlInN interface and AlGaN surface with atomic-level flatness were achieved in our DBR structure, indicating atomic-layer control during the MOCVD growth of the high-Al-composition AlGaN-based DBR structure. Additionally, an atomic force microscopy (AFM) image of the DBR surface morphology is shown in Fig. 3. The occurrence of step terminations in the AFM image indicates a step-flow growth mode. The root-mean-square roughness of the final Al 0.5 Ga 0.5 N surface obtained from AFM image was 0.31 nm over an area of 2 × 2 μ m 2 .
The composition depth profiles of the DBR structure were measured by secondary-ion mass spectrometry (SIMS). A perfectly periodic structure was observed in the depth profiles of the group III (In, Ga and Al) atom fraction, as shown in Fig. 4. The Al atomic fractions were determined to be 50% and 96.7% for AlGaN and AlInN, respectively. In the AlInGaN layer, In atoms were stable and uniformly distributed, with the atomic fraction of 2%, and the Al atomic fraction gradually increased from 75% to 95% along the growth direction. Figure 4 shows that the Al atomic fraction was 50% in the AlGaN template.
High-resolution X-ray diffraction (HRXRD) was used to investigate the structural characteristics of the AlGaN/AlInN/AlInGaN DBR. Figure 5 shows the reciprocal space map of the asymmetric (105) reflections. Multi-order satellite peaks that originated from the periodical tri-layer structures are clearly observed, indicating that the DBR structure exhibited good periodicity and clear interfaces, as observed by TEM. The maximum reflections in three envelopes of these satellites correspond to the lattice parameters of AlInN, AlInGaN and AlGaN, respectively. The Al compositions in the AlGaN and AlInN layers, as calculated according to the positions of the maximum reflections in the asymmetric (105) plane, were 50% and 96.5%, respectively, in good agreement with the SIMS results. The reflections of AlGaN, AlInN, and AlInGaN layers in the DBR were aligned perfectly in the Q y direction. A very slight difference in the maximum reflection between the AlGaN template and the DBR structure in the Q x direction indicates that a slight strain relaxation occurred during the growth of the DBR. The  XRD results indicate that the entire DBR structure had the same in-plane lattice parameter and was grown coherently on the thick Al 0.5 Ga 0.5 N template despite the relatively large lattice mismatch between the Al 0.5 Ga 0.5 N and Al 0.965 In 0.035 N layers. Furthermore, the separation of successive satellite peaks gives a DBR periodicity of 54.6 nm, in very good agreement with the 54.5 nm DBR periodicity value obtained from the TEM image.
The X-ray reflectivity results are shown in Fig. 6. The pronounced periodic oscillations indicate well-defined interfaces and uniform layer thicknesses. The fitting of the X-ray reflection spectrum reveals that the average thickness of the periodic AlGaN/AlInN/AlInGaN tri-layer structures and the interface roughness were 54.4 nm and 0.8 nm, respectively. The thickness obtained by the fitting is in good agreement with the values obtained from the cross-sectional TEM image in Fig. 1. However, the interface roughness calculated from the X-ray reflection spectrum is larger than the surface roughness of 0.31 nm measured by AFM. Actually, the measured X-ray reflection curve declines slower than the fitting curve at larger incident angles, indicating that actual interface roughness should be smaller than the value extracted by fitting. In general, the interface is smooth because the roughness measured by AFM is on the atomic scale. Figure 7 shows the experimental and calculated reflection spectra for the 20-period AlGaN/AlInN/AlInGaN DBR structure. Reflectivity was measured at room temperature near the normal incidence using a UV-vis spectrophotometer. The DBR exhibits a peak reflectivity of 86% and a stopband full-width at half-maximum (FWHM) of 16 nm, with the centre wavelength of 274 nm located in the solar-blind UV region, as shown by the solid line in Fig. 7. Although the reflectivity reaches a record high value in the solar-blind UV region, it is much lower than   Fig. 7. The absence of interference fringes at the high-energy side of the DBR stopband observed in the measured spectrum explains this reflectivity difference. This absence indicates that an obvious residual absorption exists in the stopband because the absorption edge of Al 0.5 Ga 0.5 N with the calculated value of 268 nm is very close to the high-energy edge of the DBR stopband. This reduced reflectivity is mainly attributed to the residual absorption. Other optical scattering losses arising from the crystalline quality of the DBR can be neglected owing to the perfectly periodic structure with a flat interface, smooth surface, and uniform layer thickness. We calculated the residual absorption coefficient (at the centre wavelength of the stopband) to be approximately 7.35 × 10 3 cm −1 by fitting the experimental reflectivity spectrum, as shown by the dotted line in Fig. 7.
The structural design of the DBR could be improved further by appropriately increasing the Al content of the AlGaN layer. The AlInN/AlGaN thickness ratio could also be further optimized by taking the residual absorption into account.

Discussion
We previously fabricated Al 0.4 Ga 0.6 N/AlN periodic structures and observed that the Al content in the Al 0.4 Ga 0.6 N layer exhibited a large change along the growth direction, as measured by energy-dispersive X-ray line scanning; we attributed this change in Al content to the strain induced by the lattice mismatch between Al 0.4 Ga 0.6 N and AlN 23 . Therefore, in this study, we intentionally designed a composition-graded AlInGaN transition layer between the AlGaN and AlInN layers to accommodate the lattice mismatch between AlInN and AlGaN.
We developed a novel concept for the fabrication of AlGaN-based solar-blind UV DBRs using the tri-layer AlGaN/AlInN/AlInGaN periodic structure grown on the Al 0.5 Ga 0.5 N template. The TEM and XRD results show that a state-of-the-art DBR structure with atomic-level-flatness interfaces was achieved. The DBR exhibits a peak reflectivity of 86% at the centre wavelength of 274 nm and a stopband with a FWHM of 16 nm. Owing to the residual absorption from the AlGaN layer, this peak reflectivity is lower than the theoretically calculated value. We consider that this tri-layer period structure is more efficient for the realization of high-quality DUV DBRs under the premise that the fabrication method does not increase the difficulty associated with the growth of the  DBRs. The advantage of this structure is that it is favourable for controlling the strain of multi-period heterostructures and for fabricating better DUV DBRs.

Methods
Fabrication. AlGaN-based solar-blind UV DBRs were fabricated on the AlGaN template on 2 inch c-plane sapphire substrates via MOCVD. Trimethylgallium, trimethylaluminium, trimethylindium and NH 3 were used as Ga, Al, In and N precursors, respectively. H 2 and N 2 were used as carrier gases. After the sapphire substrates were treated under ambient H 2 at 1050 °C for 5 min, the temperature was lowered to 600 °C to grow the 20-nm-thick AlN nuclear layer; the temperature was then increased to 1100 °C to grow the 990-nm-thick Al 0.5 Ga 0.5 N template. Finally, the 20-period AlGaN/AlInN/AlInGaN films as an integrated tri-layer DBR structure with AlGaN as the last layer were deposited periodically on the Al 0.5 Ga 0.5 N template at a constant temperature of 1000 °C. The growth pressure in the reactor chamber was maintained at 50 Torr. During the growth of the tri-layer DBR periodical structure, no ramping process was needed 24,25 . Measurements. The structural characteristics of the DBR were investigated using HRXRD (Philips, Panalytical X'pert, Cu Kα radiation), point-by-point-corrected SIMS, AFM, and TEM. The TEM samples were prepared using the in situ focused ion beam (FIB) lift-out technique on the FEI Helios 650 dual-beam FIB scanning electron microscope and were subsequently capped with a carbon layer followed by a locally capped protective Pt layer prior to milling. Images were collected using an FEI Tecnai TF-20 FEG transmission electron microscope operated at 200 kV. The reflectance spectra were measured at room temperature near normal incidence using a UV-vis spectrophotometer.