Researchers reveal secrets of metalorganic preflow for AlN MOCVD process
Figures a), b) and c) show electron energy loss spectroscopy maps of the Al L2,3 (green), and nitrogen (blue), oxygen (red), and carbon (yellow) K-edge at the AlN/sapphire interface and the schematics of film polarity and growth mechanism of Sample 1, 2 and 3
AlN is commonly used as template for UV optical and ultrawide-bandgap electronic devices on commercially-viable c-plane sapphire substrates. Achieving Al-polar AlN films depends primarily on the initial growth stage conditions including the substrate pre-treatments by NH3 preflow (i.e. nitridation) or trimethylaluminum (TMAl) preflow. Both pretreatment methods have been shown to lead to the Al-polarity, which is preferred over the N-polarity due to better crystalline quality and smooth surface.
While the working principle of the nitridation process has been reported and well accepted, that of the TMAl preflow was still unclear. Some researchers proposed a hypothetical Al-layer model where a number of Al layers from TMAl are formed on sapphire, leading to the Al polarity. However, there hasn't been any experimental evidence to support the hypothesis. Therefore, it was puzzling exactly how the TMAl preflow can lead to the Al-polarity and how it might impact the growth mode.
Researchers from King Abdullah University of Science and Technology (KAUST), Georgia Tech and Qatar University led by KAUST's Xiaohang Li, have conducted detailed electron microscopy analyses of the impurities, in particular, at the AlN/sapphire interface where the AlN nucleation layer was grown. They found that carbon played a primary role in determining the polarity and growth mode of the subsequent AlN film.
In the study, the MOCVD process occurred on sapphire substrates. TMAl and NH3 were used as III and N precursors with H2 as the carrier gas. Each growth process started with a TMAl preflow, with three increasing preflow doses ( Sample 1, Sample 2, and Sample 3, respectively). Subsequently, the AlN growth started with a 15 nm nucleation layer followed by a 3 µm AlN template layer.
Sample 1 does not have any preflow. The electron energy loss spectroscopy (EELS) maps, as shown in figure (a), indicate uniformly distributed Al and N elements without the presence of carbon or oxygen in the NL. But the AlN template tends to have mixed-polarity thus rough surface.
For Sample 2 with a moderate level of the preflow dose, figure (b) shows a relatively uniform distribution of oxygen and nitrogen elements but a slight accumulation of carbon at the AlN/sapphire interface, consistent with a slight reduction of Al concentration. These maps suggest the presence of scattered carbon-rich clusters near the NL/sapphire interface. The resulting AlN template has Al-polarity with smooth surface, indicating Al polarity.
Sample 3 has the largest preflow dose (i.e. overdose). As shown in figure (c), large carbon-rich clusters were formed at the AlN/sapphire interface, which attracted surrounding oxygen simultaneously. While the AlN template possessed Al polarity, the carbon-rich clusters at the NL acted as a mask. This led to a 3D growth mode throughout the subsequent template growth and hence, rough surface. The polarity of the three samples was confirmed by KOH etching.
"By using a high-resolution transmission electron microscope to investigate the element distribution, mainly aluminium, nitrogen, carbon, and oxygen, at the AlN/sapphire interface, we experimentally showed the strong correlation between the formation of carbon clustering at the AlN/sapphire interface and the polarity and growth mode of AlN films," said Haiding Sun, a postodoc in Xiaohang Li's group.
"The key takeaway is that carbon from the preflow captured oxyygen on sapphire surface, preventing formation of AlxOyNz and thus N-polar AlN. The capture was due to the low formation energy for the CN-ON, or a kind of complex, such as Al-C-O complex. But overdose can lead to large C-cluster masks creating 3D growth mode" Sun added. "These results show the importance of the precise control of the carbon presence on the surface prior to the AlN epitaxy."
The researchers plan to further investigate the correlation between TMAl preflow and the crystal quality of the grown AlN films. Specifically, different types of disclocation generation in the films are of interest. Further results will be reported.
This collaborative work, which received financial support from KAUST, GCC Research Program, DARPA, and NSF is detailed in the paper: 'Influence of TMAl preflow on AlN epitaxy on sapphire', from Applied Physics Letters 110 (20) 192106 (2017) published on May 12, 2017.
