Introduction

Indium gallium nitride (InGaN), one of the compound semiconductor materials, has been applied extensively in optoelectronics. For instance, InGaN quantum dots (QDs) have excellent characteristics for optoelectronic devices, such as light emitting diodes (LEDs) [1,2,3], infrared photodetectors [4,5,6], and laser diodes [7,8,9]. The most commonly used mechanism for the growth of self-assembly QDs is Stranski–Krastanov (SK) mode [10,11,12] through molecular beam epitaxy (MBE) [13, 14] and metal–organic chemical deposition (MOCVD) [15,16,17]. However, the SK mode has drawbacks as the growth of InGaN QDs triggered by layer strain, which can produce threading dislocations or defects, and uncontrollable density [10, 18]. On the other hand, the growth of crystal InGaN QDs with indium-rich composition and low-dislocation density is still challenging to broaden the emissions spectra from the visible red to the near-infrared regions [19]. Droplet epitaxy (DE) could be an alternative method to solve this problem for fabricating semiconductors' QDs [20, 21]. This is because DE technique has several advantages for the crystal growth of QDs, such as the strain-independent materials, lower temperature process, lattice-mismatched substrates, and the QDs growth without a wetting layer [22]. Some fundamental aspects in controlling the dimensions and density of GaN and InN QDs by DE have been investigated, such as the growth temperature, droplet deposition time, and the duration of nitridation [23, 24]. So far, there seems little or no report on the formation of InGaN QDs via the DE technique.

In this work, InGaN QDs have been fabricated on Si (111) substrate by forming In–Ga alloy droplets and then the nitridation process in the plasma-assisted molecular beam epitaxy (PA-MBE) system. The growth parameters of substrate temperature, deposition time of In–Ga droplets, and plasma nitridation duration have been adjusted to investigate the growth mechanism of InGaN QDs by DE. Some characterization tools were employed to analyze the morphology, chemical composition, crystallinity, and optical properties of InGaN QDs. This report demonstrates the growth of InGaN QDs by DE and the potential applications in long-wavelength optoelectronic devices.

Methods

Substrate preparation

A 2-inch silicon (111) wafer as a substrate was cleaned with acetone for 10 min and rinsed with deionized (DI) water in the ultrasonic oscillator. After that, the substrate was soaked in 10% hydrofluoric acid for 10 min and washed with DI water. Finally, the substrate was blown-dry with nitrogen gas and then immediately put into the load-lock chamber of our ULVAC PA-MBE system.

InGaN QDs growth

Before the growth, the base pressure of growth chamber in PA-MBE system was ~ 4 × 107 Pa. The Si substrate was thermally cleaned in the growth chamber at 600 °C for 10 min. The substrate was conducted the pre-nitridation process under 500 W RF plasma with 0.8 sccm high purity nitrogen for 30 min. Two source K-cells, filled with indium (6N) and gallium (6N), respectively, were maintained at the temperature of 800 °C. The beam equivalent pressures of In and Ga K-cells were 1.13 × 105 and 8 × 106 Pa, respectively. Different substrate temperatures (350, 400, and 450 °C) were set before the growth. For droplet epitaxy, the K-cells supplied In and Ga droplets for 30, 45, and 60 s. And then, the plasma nitridation process for those droplets on the surface for 0, 15, 30, and 45 min, respectively. The detailed parameters for the droplet epitaxy of InGaN QDs are listed in Table 1. The growth parameters include the substrate temperature of growth, the time of In–Ga droplet deposition, and the duration of plasma nitridation.

Table 1 Sample growth parameters for droplet epitaxy of InGaN QDs
Full size table

InGaN QDs characterizations

In the growth chamber of MBE, an in-situ 20 kV reflection high energy electron diffraction (RHEED) was used to monitor the surface condition of samples and the growth of InGaN QDs. After the growth, the surface morphology was observed by JEOL JSM-7000F field emission scanning electron microscopy (FE-SEM). Surface roughness and local contact voltage as a work function of the material were measured by Nanosurf C3000, including atomic force microscopy (AFM) and Kevin probe force microscopy (KPFM). The chemical bonding state and composition of InGaN QDs were determined by VGS Thermo K-alpha X-ray photoelectron spectroscopy (XPS). The microstructure of InGaN QDs was identified by JEOL JEM-3010 transmission electron microscopy (TEM) with energy-dispersive X-ray spectroscopy (EDX). The photoluminescence (PL) spectra of the InGaN QDs were obtained in the range of 530 and 900 nm, equipped with a 500 μW laser diode in the wavelength of 450 nm.

Results and discussion

During the growth of InGaN QDs, there are three main stages: pre-nitridation of Si substrate, In–Ga droplet deposition, and the formation of InGaN QDs. The schematic of the process is illustrated in Fig. 1a, which corresponds with the RHEED patterns in Fig. 1b–d, respectively. After the pre-nitridation process in Fig. 1b, the streaky RHEED pattern with Kikuchi lines indicates the flat surface formed with a crystalline structure [10, 14, 25]. A very thin SiNx layer was formed because nitrogen atoms absorbed on the Si surface, which will also be reported by the observation of HR-TEM. After In–Ga droplet deposition at the growth temperature, the RHEED pattern in Fig. 1c became foggy. It reveals that the deposition of In–Ga alloy droplets is an amorphous structure [25]. In and Ga atoms from K-cells impinged, adsorbed, and migrated on the surface. After following exposure to nitrogen plasma, the In–Ga droplets grew and completely crystallized into InGaN QDs. Meanwhile, the RHEED pattern changed from foggy to ring-type, as shown in Fig. 1d, which performed the formation of polycrystalline InGaN QDs on the SiNx/Si surface [24].

Fig. 1
figure 1

a Illustration of InGaN QDs growth on the Si (111) substrate via droplet epitaxy, b RHEED pattern of substrates after pre-nitridation at 600 °C for 30 min, c RHEED pattern after In–Ga deposition, and d RHEED pattern of InGaN QDs for B3 sample

Full size image

FE-SEM was used to observe InGaN QDs and the surface morphologies of all samples, shown in Fig. 2. The average lateral dimension and density of InGaN QDs on Si can be calculated from the images, which are summarized in Table 2. Three different growth parameters will be discussed. First, Fig. 2a, b shows FE-SEM images of A1 and A2 samples, respectively. The higher growth temperature increased the average diameter of InGaN QDs from 13.3 ± 3.0 to 28.7 ± 13.2 nm, and the dot density decreased dramatically from 1.33 × 1011 to 1.66 × 1010 cm−2. These can be attributed to the improved migration length and aggregation of adatoms to form bigger dots at higher growth substrate temperatures [14]. The evaporation of indium from the substrate at higher growth temperature also occurred during the growth, which could significantly decrease the dot density. In addition, the higher standard deviation of dot diameter (28.7 ± 13.2 nm) was observed for sample A2, grown at 450 °C. It is because of the Ostwald ripening phenomenon, which made the In–Ga droplets bigger and inhomogeneous InGaN QDs after nitridation. Both nanostructures of InGaN quantum dots and rings can be observed in this sample, shown in Fig. 2b. Second, when the duration of plasma nitridation increased from 0 to 45 min, images are shown in Fig. 2c for sample B1 and in Fig. 2d for sample B3, respectively. The average diameter of InGaN QDs increased from 14.7 ± 2.2 to 24.8 ± 8.2 nm. For sample B1 (without nitridation), In–Ga alloy droplets on the surface performed in a round shape. After the nitridation process for 45 min, sample B3, InGaN QDs in irregular shapes were observed, and the standard deviation of dot diameter increased. During the nitridation process, the nearby small dots could diffuse together to form bigger dots. The longer nitriding time supplied more active nitrogen radicals to bond with In and Ga adatoms to form InGaN QDs [20]. The dot density reduced from 1.11 × 1011 to 1.0 × 1011 dots/cm2. Third, the different deposition time of In–Ga droplet was investigated. In Fig. 2e, f, the deposition time from 45 to 60 s made the dot diameter get bigger from 22.4 ± 6.3 to 23.5 ± 8.6 nm, and also increased dot densities from 8.51 × 1010 to 8.96 × 1010 dots/cm2. The longer In–Ga deposition time supplied more atoms to get a larger dot size and higher density after the growth. The influence of In–Ga deposition time seems not prominent at fixed growth temperature and nitridation time.

Fig. 2 
figure 2

FE-SEM images of InGaN QDs on Si (111) substrate: a A1, b A2, d B3, e C1 and f C2; In–Ga droplets on Si(111) substrate: c B1

Full size image
Table 2 Average diameter and density of InGaN QDs or In–Ga droplets
Full size table

Figure 3 shows AFM and KPFM images of InGaN QDs on the Si substrate with different growth parameters. The scan area of AFM measurement was set as 1 µm × 1 µm. Besides, KPFM was used to characterize the local contact potential difference (LCPD) as a work function between the AFM tip and the sample surfaces of InGaN QDs [26, 27]. The InGaN QDs can be observed in the AFM images in Fig. 3, and the average surface roughness can be calculated in Table 3. The size of InGaN QDs and surface roughness increased with higher growth temperature, which is consistent with the results of SEM observation. InGaN QDs grown at 350 °C had a smaller average diameter and performed a smaller surface roughness of 1.27 nm, as presented in Fig. 3a. When the growth temperature was increased to 450 °C, the diameter of InGaN QDs became three times bigger, and the average surface roughness increased to 4.67 nm, shown in Fig. 3b. Meanwhile, the surface potential decreased from 397.02 to 243.40 mV. As the nitridation duration increased from 0 to 45 min, shown in Fig. 3c, d, bigger dots were obtained, and surface roughness increased from 2.57 to 3.82 nm. Interestingly, LCPD values of B1 and B3 decreased dramatically from 680.11 to 365.9 mV in Fig. 3g, h. It can evidence that the nitridation process of the DE technique made In–Ga alloy droplets transform to InGaN QDs. Moreover, as the growth at higher temperature or a prolonged nitridation time, the average LCPD values decreased, which means the surface potential decreases with decreasing indium content of InGaN [28]. Finally, the deposition time of In–Ga droplets increased from 45 to 60 s, shown in Fig. 3e, f. The average roughness increased from 3.78 to 4.80 nm, which is consistent with the observation of SEM images.

Fig. 3
figure 3

AFM images: a A1, b A2, c B1, d B3, e C1, and f C2; KPFM images: g B1, and h B3

Full size image
Table 3 Root mean square (RMS) of AFM and local contact potential difference (LCPD) of KPFM for all samples
Full size table

The surface chemical condition of InGaN QDs was measured by XPS, and the chemical composition of InGaN can be further analyzed by the fitting software. The de-convoluted XPS spectra for In 3d5/2 are presented in Fig. 4a with the fitting area of In–In, In–N, and In–O bonding for samples B1, B2, and B3. Similarly, Fig. 4b shows the spectra of Ga 2p with the fitting area of Ga–Ga, Ga–N, and Ga–O bonding. For sample B1 (without nitridation), the peak positions of binding energy for In–In, In–N, In–O, Ga–N, and Ga–O are around 443, 444.2, 445.2, 1118, and 1119 eV. The fitting areas of In 3d5/2 and Ga 2p with the binding for In–In, In–N, In–O, Ga–Ga, Ga–N, and Ga–O are 4, 34, 62, and 0, 70, 30%, respectively. The high content of In-O and Ga-O came from the absorption of oxygen and oxidation on the surface of In–Ga droplets for B1. When the nitridation duration increased for samples B2 and B3, more impinging nitrogen radicals with In–Ga adatoms and nitridation reaction made Ga–N and In–N bonds dominate [29]. Sample B3 can obtain 100% Ga–N bond and without any Ga–O bond. In short, XPS data in Fig. 4 also can evidence the nitridation reaction of InGaN QDs by droplet epitaxial growth. However, the In–O bonds still existed in B2 and B3 as 22 and 25%, respectively. It is because of the difference in formation enthalpy between GaN and InN, that could result in the segregation of indium clusters after droplet epitaxy [30].

Fig. 4
figure 4

De-convoluted a In 3d5/2 and b Ga 2p XPS spectra for samples B1, B2, and B3

Full size image

The XPS data for all samples is summarized in Table 4. The fitting area can be used to roughly estimate the average chemical composition of InGaN on the surface without considering the phase separation effect [31]. To compare the parameter of growth temperature for samples A1 and A2, higher growth temperatures made indium composition redue from 82 to 60%, but increased gallium from 18 to 40%. At higher growth temperature, Ga atoms are more accessible to absorb, diffuse and react with nitrogen than In atoms. However, the phase separation, InN decomposition, and In desorption can occur at higher temperatures [32]. In short, the substrate temperature is also an important parameter for InGaN QDs with high-indium content by droplet epitaxy.

Table 4 The peak position, fitting area, and average surface chemical composition for all samples
Full size table

InGaN QDs have been observed by TEM after the preparation of sample B3 by focused ion beam technique. In Fig. 5a of the cross-sectional TEM image, self-assembled InGaN QDs have been grown on the surface of Si substrate by the growth mode of droplet epitaxy. Figure 5b shows the enlargement of the red rectangular area in Fig. 5a by the high-resolution TEM image. Three InGaN nanodots were observed, and the one in the center has a height of 11.58 nm. The dot at the lower right corner of Fig. 5b obviously performs a single crystal structure of InGaN. A very thin amorphous SiNx layer between the Si substrate and InGaN QDs can be observed, due to the pre-nitridation treatment on the Si substrate [33]. Moreover, the chemical composition analysis was conducted by EDX spectral mapping on elements In, Ga, and N, presented in Fig. 5c, respectively, which also could confirm the droplet epitaxial growth of InGaN QDs.

Fig. 5
figure 5

a Cross-sectional TEM image of InGaN QDs grown on the Si (111) substrate for sample B3, b High-resolution TEM image for the red rectangular area, and c EDX mapping on the elements of indium, gallium, and nitrogen

Full size image

The optical properties of uncapped InGaN QDs for samples A1, A2, and B3 were characterized by PL spectroscopy measured at the temperature of 80 K. Figure 6 shows three major emission peaks around the wavelength of 715, 795 and 857 nm, and the emissions of InGaN QDs are in the wavelength of visible red and infrared region. The broad emission peaks indicated the inhomogeneous size, composition gradient, or random distribution of InGaN QDs [34]. The spiky PL peaks could be the noise from InGaN QDs without capping layer, which attributed to the surface recombination of carriers.

Fig. 6
figure 6

PL spectra at the temperature of 80 K for A1, A2, and B3 samples

Full size image

The bandgap of InxGa1−xN alloy is in the range of 0.7–3.4 eV, depending on the composition of indium [35]. The emission spectra of InGaN QDs in the wavelength of visible red and infrared region could imply the high-indium composition of InGaN QDs. Among all the samples, B3, the sample with longer nitridation duration, exhibited the highest peak intensity of PL spectra. It indicates that the duration of nitridation could be the most important parameter for the growth of InGaN QDs via droplet epitaxy.

Conclusions

The first investigation for the droplet epitaxial growth of InGaN QDs on Si (111) substrate was proposed by using the PA-MBE system. The growth of InGaN QDs was signed with the ring-type RHEED pattern, indicating the transformation from the amorphous In–Ga droplets to the polycrystalline InGaN structure. Moreover, high-resolution TEM data also proved the crystal growth of uncapped InGaN QDs. Three parameters, substrate temperature, duration of nitridation, and In–Ga deposition time were conducted to study the growth of InGaN QDs. During the droplet epitaxial growth, different substrate temperatures influenced adatoms’ adsorption, desorption, and surface diffusion to control the average size and density of InGaN QDs. The higher density of 1.33 × 1011 cm−2 and smaller size of 13.3 ± 3.0 nm can be obtained at the lower growth temperature of 350 °C. Then, the duration of plasma nitridation had less impact on the density of InGaN QDs, but it was an essential parameter in the formation of crystal InGaN QDs. The round-shape In–Ga alloy droplets became polygon of InGaN QDs. As the duration of nitridation increased, the surface chemical compositions of In–N and Ga–N bonds increased, and intensity of photoluminescence enhanced. It could be due to the crystal-quality improvement of InGaN QDs. The emissions of InGaN QDs were in wavelength of the visible red and infrared region, indicating the formation of the high-indium composition of InGaN QDs via droplet epitaxy technique.