GRAPHENE

Demonstration of epitaxial growth of strain-relaxed GaN films on graphene/SiC substrates for long wavelength light-emitting diodes


To keep away from the inevitable air pollution, mechanical harm, and measurement limitation of graphene in the course of the switch course of, graphene was immediately ready on Si-face 4H-SiC substrates by thermal decomposition. Determine 1a exhibits floor morphology of the ready graphene on SiC substrate measured by the atomic drive microscopy (AFM). The AFM picture presents a step-like floor morphology with a constant path, uniform width and peak. The measured terrace width and step peak are about 5–10 μm and 10–15 nm, respectively. To characterize the uniformity of the graphene on SiC substrate, we measured Raman spectra at 5 completely different positions on a 2″ diameter graphene/SiC substrate, as proven in Fig. 1b. The inset shows the {photograph} of the two″ graphene/SiC substrate and the measured positions. The attribute G peak (~1582 cm−1) and 2D peak (~2700 cm−1) of graphene may be seen in Raman spectra of 5 measured positions, which signifies that the ready graphene on SiC substrate by thermal decomposition has good uniformity19,20. As well as, we calculated the depth ratio of the attribute 2D-to-G peak (I2D/IG) of graphene, and the depth ratio of the defect-related D peak (~1350 cm−1) to the attribute G peak (ID/IG) of that. The outcomes present that the ready graphene on SiC substrate is the multilayer one with good high quality (see Fig. S1)21.

Fig. 1
figure1

Characterizations of graphene/SiC substrates. a AFM picture of graphene/SiC substrate, exhibiting the step-like construction. b Raman spectra of 5 measurement positions on 2″ graphene/SiC substrate. The inset exhibits the {photograph} of as-grown 2″ graphene/SiC substrate and the measured positions

GaN movies have been grown on graphene/SiC substrates utilizing MOCVD. The shortage of dangling bonds on graphene floor is just not conducive to the nucleation development of GaN17. The as-grown GaN on graphene can’t type a steady movie, and the floor morphology presents a discrete island-like distribution with unbiased nucleation, as proven in Fig. 2a. In an effort to improve the nucleation skill of nitrides on graphene, we carried out nitrogen-plasma pre-treatment on graphene earlier than the epitaxial development, and optimized the buffer layer within the epitaxial construction (see Fig. S2). Based mostly on the above efforts, the continual and easy GaN movies have been obtained on graphene through the use of low- and high-temperature AlN buffers, as proven in Fig. 2b.

Fig. 2
figure2

Floor morphology of GaN on graphene/SiC. SEM photos of GaN on: a untreated graphene, and b nitrogen-plasma-treated graphene

Cross-sectional high-angle annular dark-field STEM (HAADF-STEM) measurements have been carried out on the as-grown GaN on untreated graphene. As proven in Fig. 3a, the epitaxial construction consists of GaN islands, AlN buffer and graphene/SiC substrate on untreated graphene, verifying that GaN on untreated graphene exhibits the discrete islands development. Determine 3b exhibits the HAADF-STEM picture of the interface area marked with a pink rectangular body in Fig. 3a. It may be seen that the AlN buffer grown on graphene is just not fully merged, and there are not any GaN nucleation islands above the AlN buffer, indicating that GaN nucleation islands are distributed. Determine 3c exhibits the cross-sectional high-resolution transmission electron microscopy (HRTEM) picture of the interface area marked with a blue rectangular body in Fig. 3b, and the multilayer graphene on the interface may be noticed clearly. The inset of Fig. 3c exhibits the chosen space electron diffraction (SAED) sample of the AlN buffer within the white sq. area in Fig. 3c. This consequence signifies that the AlN buffer grown on graphene is polycrystalline. Multilayer graphene can successfully display the lattice potential area from SiC substrate, ensuing within the forming of the polycrystalline AlN buffer22. The SEM and X-ray diffraction (XRD) measurement outcomes additionally verify that AlN buffer is polycrystalline (see Fig. S3). The interface elemental elements have been analyzed by power dispersive spectroscopy (EDS) mapping. Determine 3d exhibits the EDS mapping photos of various components at AlN/graphene/SiC interface. The distributions of C, Si, Al, and N components verify that the graphene nonetheless steadily exists after AlN epitaxy, indicating that the graphene is just not broken below excessive temperature and hydrogen ambiance of MOCVD system, and nonetheless maintains its integrity.

Fig. 3
figure3

HAADF-STEM evaluation of GaN on untreated graphene/SiC. a Cross-sectional HAADF-STEM picture of GaN on untreated graphene. b Cross-sectional HAADF-STEM picture of the interface space marked with a pink rectangle body in (a). c Cross-sectional HRTEM picture of the interface space marked with a blue rectangle body in (b). The multilayer graphene on the interface is clearly seen. The inset in (c) exhibits the SAED sample of AlN within the white sq. area. d EDS mappings of elemental C, Si, Al, and N on the interface of AlN/graphene/SiC

The Raman spectra in Fig. 4a present that the D peak depth of nitrogen-plasma-treated graphene is significantly elevated in contrast with untreated one, which signifies that defects are launched on graphene floor by nitrogen-plasma pre-treatment. After that remedy, the ID/IG ratio of the graphene in Raman spectra will increase from 0.3 to 0.5. The defect density in graphene may be calculated based mostly on the ID/IG ratio, and it will increase from 9.0 × 1010 cm−2 to 2.0 × 1011 cm−223. In an effort to additional research the impact of nitrogen-plasma pre-treatment, X-ray photoelectron spectroscopy (XPS) measurements have been carried out for untreated and nitrogen-plasma-treated graphene. Determine 4b exhibits the XPS spectrum of C 1s peak of untreated graphene, which consists of Si–C bond (~283.2 eV) from SiC substrate and sp2C–sp2C bond (~284.6 eV) from graphene24. After nitrogen-plasma pre-treatment, two completely different sorts of C–N bonding configurations appeared in C 1s peak, equivalent to N–sp2C bond (~285.5 eV) and N–sp3C bond (~286.5 eV), as proven in Fig. 4c25. Because the pyrrolic N atom in N–sp3C bond is extra reactive than the pyridinic N atom in N–sp2C bond11,18, N–sp3C bond can present extra nucleation websites for subsequent epitaxial development of GaN movies.

Fig. 4
figure4

Raman and XPS analyses of graphene earlier than and after nitrogen-plasma pre-treatment. a Raman spectra of untreated graphene (blue) and nitrogen-plasma-treated graphene (pink). The inexperienced and black diamond blocks correspond to the attribute peaks of SiC and graphene, respectively. The orange dashed area exhibits a major enhance within the depth of D peak. XPS spectra with C 1s of: b untreated graphene, and c nitrogen-plasma-treated graphene. XPS outcomes present that nitrogen-plasma pre-treatment introduces C–N bonds on graphene floor: N–sp2C bonds (~285.5 eV) and N–sp3C bonds (~286.5 eV)

Determine 5a exhibits the cross-sectional HAADF-STEM picture of GaN on nitrogen-plasma-treated graphene. The epitaxial construction composed of GaN movie, AlN buffer and graphene/SiC substrate may be noticed clearly. Determine 5b exhibits the cross-sectional HAADF-STEM picture of the interface area marked with a pink rectangular body in Fig. 5a. It’s value noting that a part of graphene was etched after nitrogen-plasma pre-treatment. Amongst them, the darkish half indicated by the pink arrow is the unetched graphene area, and the sunshine half indicated by the blue arrow is the etched graphene area. The SAED sample in Fig. 5c signifies that the AlN buffer on the unetched graphene area continues to be polycrystalline, whereas the AlN buffer on the etched graphene area exhibits a single crystalline with hexagonal wurtzite construction, as proven in Fig. 5d. Single crystalline AlN performs a dominant function in the course of the subsequent epitaxial development, which facilitates the epitaxial development of GaN movies. Determine 5e exhibits the built-in differential part distinction (iDPC) STEM picture of the GaN on nitrogen-plasma-treated graphene. The association of Ga and N atoms signifies that the GaN movie grown on nitrogen-plasma-treated graphene is Ga polarity.

Fig. 5
figure5

HAADF-STEM evaluation of GaN on nitrogen-plasma-treated graphene/SiC. a Cross-sectional HAADF-STEM picture of GaN on nitrogen-plasma-treated graphene. b Cross-sectional HAADF-STEM picture of the interface space marked with a pink rectangle body in (a). The pink arrow factors to website the place the darkish a part of the unetched graphene, whereas the blue arrow factors to website the place the sunshine a part of the etched graphene. c, d SAED patterns of AlN on the graphene/SiC marked with a white rectangular body in (b). e iDPC-STEM picture of GaN grown on nitrogen-plasma-treated graphene/SiC. The atomic association of Ga and N atoms confirms the Ga-polarity for the as-grown GaN

To find out the macroscopic orientation and crystal construction of GaN movies on nitrogen-plasma-treated graphene, we additionally carried out XRD measurements on GaN movies. The XRD 2θ and φ scan outcomes present that GaN movie on graphene is a well-aligned single crystal with hexagonal wurtzite construction (see Fig. S4). The development of GaN nucleation on graphene not solely reduces the floor roughness of overgrown GaN, but additionally improves its crystalline high quality, as proven in Fig. 6a, b. The total-width at half-maximum (FWHM) of (0002) and (10(bar 1)2) planes rocking curve of GaN epilayer are significantly diminished from 1260 and 1440 arcsec to 232 and 290 arcsec, respectively. The crystalline high quality of GaN movies on graphene is corresponding to that of GaN movies grown immediately on different standard international substrates, comparable to Si and sapphire (see Desk S1).

Fig. 6
figure6

Characterizations of GaN movies on graphene/SiC. X-ray rocking curves of: a (0002), and b (10(bar 1)2) planes for two μm thick GaN movies grown on untreated graphene/SiC and nitrogen-plasma-treated graphene/SiC substrates. c Raman spectra of GaN movies epitaxially grown on graphene (pink) and immediately on SiC substrate (black). The inexperienced dashed line is the E2 (excessive) phonon frequency of GaN bulk materials below stress-free state

The stress in as-grown GaN movies is additional evaluated by Raman spectroscopy. The E2 (excessive) phonon mode in Raman spectra is delicate to the biaxial pressure of GaN movies, and due to this fact it may be used to judge the biaxial stress in III-nitrides26. For stress-free GaN bulk supplies, its E2 (excessive) phonon frequency is 568 cm−127. Determine 6c exhibits that the E2 (excessive) phonon frequencies of GaN movies grown on graphene and immediately on SiC are each lower than 568 cm−1, indicating that each GaN movies are below tensile stress state28. Whereas, the E2 (excessive) phonon frequency of GaN on nitrogen-plasma-treated graphene is nearer to that of stress-free GaN, which is 567.9 cm−1. The particular biaxial stress worth σ of GaN epilayers may be calculated by the system: σ = Δω/κ, the place κ is the stress coefficient, Δω is the Raman frequency shift relative to the E2 (excessive) phonon frequency of stress-free GaN movies. Right here, κ = −3.4 cm−1 GPa−1 is adopted for the calculation29. The calculated residual tensile stress worth for the GaN movies on graphene is simply 0.03 GPa, which is far decrease than that of GaN movies grown immediately on SiC substrate (0.74 GPa). This consequence exhibits that the residual stress in GaN movies may be considerably diminished by inserting graphene.

Based mostly on the above outcomes, we suggest a development mannequin to clarify the expansion mechanism of GaN movies on graphene. As proven in Fig. 7a, AlN nucleation islands grown immediately on untreated graphene exhibit a random in-plane orientation. After nitrogen-plasma pre-treatment on graphene, a part of its floor space is etched, as proven in Fig. 7c. On the identical time, C–N-related dangling bonds are fashioned on graphene floor. In Fig. 7d, the orientation of AlN on etched graphene area continues the orientation of SiC substrate, exhibiting a single crystalline c-axis orientation development. The c-axis-oriented AlN nucleation islands step by step occupy the dominant function, and a single crystal AlN layer with the identical orientation is fashioned by lateral epitaxy. The following GaN movies are grown on the AlN layer, as proven in Fig. 7e. From the cross-sectional HAADF-STEM picture, we are able to see the grain boundaries generated in the course of the lateral merging of AlN above completely different areas of graphene, as proven in Fig. 7f, which additional verifies the rationality of the proposed development mannequin.

Fig. 7
figure7

Schematic diagram of the expansion mannequin of GaN movies on nitrogen-plasma-treated graphene. a AlN nucleation islands on untreated graphene. The nucleation orientation exhibits a random in-plane orientation. b Direct development of graphene on SiC substrate. c As-grown graphene after nitrogen-plasma pre-treatment. The C–N-related dangling bonds have been fashioned on nitrogen-plasma-treated graphene, and the grey, pink, and blue spheres characterize the C, pyrrolic N, and pyridinic N atoms, respectively. d AlN nucleation islands on nitrogen-plasma-treated graphene. The nucleation orientation of AlN on the etched graphene area continues the orientation of SiC substrate, which is a single crystal with c-axis orientation. e Epitaxial development of steady GaN movies on AlN buffer. f Cross-sectional HAADF-STEM picture of the interface at AlN/graphene/SiC

Based mostly on above outcomes, InGaN/GaN MQWs have been grown on the as-grown GaN/graphene/SiC template. For comparability, we additionally ready InGaN/GaN MQWs with the identical construction on GaN/SiC template below the identical circumstances. The 2 samples have been grown in a single run. Determine 8a, b exhibit XRD 2θ scan spectra for InGaN/GaN MQWs, the place intense diffraction peaks from GaN epilayer and satellite tv for pc peaks from the InGaN/GaN MQWs as much as the fourth order may be noticed in each samples, indicating good crystalline high quality and sharp interfaces of InGaN/GaN MQWs. From the becoming outcomes, it may be discovered that the indium (In) content material in InGaN properly layer will increase from 25% to 29% after inserting graphene. Determine 8c exhibits the PL spectra of InGaN/GaN MQWs at a low temperature of 10 Ok, which signifies that the insertion of graphene ends in a red-shift of MQWs emission wavelength from 535 to 556 nm, and the shift is about 21 nm. Based mostly on the height positions of the PL spectra, the calculated In content material within the InGaN/GaN MQWs on graphene/SiC is 3.5% larger than that on SiC, which is near the XRD becoming outcomes (In content material elevated by 4%). Based on the Raman outcomes, it’s discovered that the stresses are significantly completely different within the GaN templates on graphene/SiC and SiC. The residual stress within the GaN template on graphene/SiC is far decrease than that on SiC. Due to this fact, we predict that the pressure rest is the principle purpose for the rise of In content material in InGaN/GaN MQWs.

Fig. 8
figure8

Characterizations of InGaN/GaN MQWs on SiC and graphene/SiC substrates, respectively. XRD 2θ-ω scans spectra of (0002) airplane for InGaN/GaN MQWs grown on: a SiC, and b graphene/SiC substrates, respectively. c PL spectra of InGaN/GaN MQWs grown on SiC and graphene/SiC substrates (T = 10 Ok)



Source link