Liquid metal-based 2D GaN printing technology. The schematic in Fig. 1a illustrates the preparation of ultrathin 2D GaN layer using N2-plasma treatment technology at room temperature. The plasma nitriding process employs an unusual glow discharge involving high current and charge densities. A potential difference is generated by applying a DC voltage between the components (cathode) and the furnace walls (anode), which ionizes the treatment gas to produce the glow discharge. The treatment gas’s positive ions speed up to the negatively connected parts and hit the surface with high kinetic energy, sputtering the surface and heating the whole parts. Nitrogen plasma is an electrically neutral aggregate composed of N+, N2+, e and N2 formed after nitrogen is ionized. In the plasma nitriding of gallium, nitrogen ions and accelerated neutral nitrogen atoms lie on the surface to construct a nitrogen-rich film that forms a nitride layer on the surface. Because the chemical reaction of nitrogen plasma is a thermodynamically stable excited state and ionic state, and the reaction activation energy is very low, it is much easier to generate GaN by the reaction of nitrogen plasma with liquid Ga than by the traditional nitriding reaction Ga (CH3)3 + NH3 → GaN + 3CH4.
The schematic setup of the system to transform the liquid Ga to 2D GaN is elucidated in Fig. 1b. The formation of 2D GaN thin films was realized by nitrogen plasma triggering surface limited nitriding reaction. Gallium droplets with plasma treated surface were placed on SiO2/Si substrate and moved on the surface with the help of scraper, during the scraping process of liquid Ga, the surface nitride produced by plasma bombardment in nitrogen environment sticks to the substrate through van der Waals force (Fig. 1c) (Detailed description of the process parameters will be given in the Methods.). The nitridation reaction and printing of Ga both are conducted at room temperature that is consistent with the current electronic device production processes. The covered substrate area can be expanded by choosing a larger droplet diameter and a longer travel distance of the scraper. The thickness of the film can be increased via frequent printing (Methods and Supplementary Fig. 2). Figure 1d revealed a large and continuous ultrathin GaN film reaching lateral dimensions more than many centimeters. Based on atomic force microscopy (AFM), the thickness of the deposited GaN layer is ~4 nm that is moderately greater than a single GaN unit cell (Fig. 1e). AFM also indicates that the printed GaN film’s surface roughness is analogous to that of the SiO2 substrate, demonstrating that the 2D GaN film has minimum cracks, holes, folds, or bubbles, reflecting conformal and homogeneous attachment. The uniform thickness in broad regions and numerous samples supports the proposed Cabrera–Mott growth mechanism, where concurrent nitrides are formed among the whole metal interface, resulting in self-limiting growth to an accurate thickness. The mentioned approach is highly reproducible for growing large-area GaN sheets because the process was terminated more than 50 times and always yielded identical, continuous, laterally extensive atomically thin GaN films with reproducible features. The GaN film was initially printed on the SiO2/Si substrate, but further tests demonstrated that the formation of uniform centimeter-scale semiconductor films on various substrates could be reproduced through the printing technology, indicating that the presented construction approach is suitable to deposit 2D GaN for several materials. Moreover, it should be pointed out that this method is also applicable to the fabrication of GaN heterostructures. In order to be compatible with silicon-based electronic technology, the 2D film printed on SiO2/Si surface should be employed for further description and device construction.
Properties of the 2D GaN. Transmission electron microscopy (TEM) can be utilized to verify the crystallographic features of printed 2D GaN. The 2D GaN films were immediately moved to a TEM grid since it was printed. A high-resolution TEM micrograph (HRTEM) and selective area electron diffraction (SAED) pattern of the GaN are presented in Fig. 2a and 2b, respectively, which confirms the crystallization of the printed GaN in the polymorph. The atomic spacing of 0.285 nm in the HRTEM image and the distance of the fringes at 0.266 nm correspond to the (100) and (002) planes of GaN, respectively, and that at 0.248 nm corresponds to the (101) plane.
The 2D GaN film’s phonon modes also change compared to those for the bulk17. Based on the Raman spectra presented in Fig. 2c, two peaks at 561 and 735 cm−1 appear in 2D printed GaN, respectively, corresponding to E1 (TO) and A1 (LO) modes in the bulk phase. Moreover, the above distinction reflects the change in GaN’s phonons modes in 2D limit, while its location at 566.2 cm−1 is blue-shifted comparing to that of the bulk, reflecting a tensile-strain state in the 2D limit18. The peak position of A1 (LO) mode corresponds to 731 cm−1, and there is a blue shift relative to the peak position of 736 cm−1 in the bulk phase. Generally speaking, the parameters influencing Raman scattering involve material size, order, internal stress and structural defects. Copwell et al.19 believe that the reduction of nano material size will lead to the movement, broadening and asymmetric peak shape. In addition, the proposed spectrum lacks various Raman properties of Ga2O3 (i.e., peaks at ~167, ~320, ~344, and ~475 cm−1), indicating that GaN is successfully constructed and Ga is quantitatively changed within plasma discharge nitriding reaction. It was observed that the Raman peak intensity for 456.6 cm−1 modes was notably decreased than that of the bulk counterpart, while the A1 Raman mode is susceptible to the free charge carrier density in graphene and 2D metal chalcogenides20–23. Due to its broader bandgap, 2D GaN can involve further trap states. The appearance of more polar 2D GaN changes the interaction between phonon and free charge carriers, obtained by the trap states, inside the 2D material, resulting in phonon self-energy renormalization. Thus, phonons are attenuated, decreasing the charge-sensitive A1 Raman mode’s intensity20–23.
X-ray photoelectron spectroscopy (XPS) is utilized to attain the printed 2D GaN’s chemical bonding states. Figures 2d and e present the spectra of Ga 2p and N 1s areas for the GaN, respectively. The doublet in the Ga 2p region corresponds to the 2p3/2 and 2p1/2 orbital of Ga, the characteristic gallium peak for Ga2O3 placed at ~20.4 eV was not seen, indicating Ga’s quantitative transformation. The principal broad N 1s peak centered at ~397.6 eV corresponds to the N 1s region compatible with the desired N 1s area presented in GaN. Energy-dispersive X-ray spectroscopy (EDS) mapping study of the resulting film composition (Supplementary Fig. 3) indicated that the deposited layer is mainly GaN with nitrogen to gallium ratio around 1, confirming the stoichiometric GaN. High-performance 2D semiconductors with a large area and good uniformity is necessary for empirical electronic device applications.
Gallium Nitride has a predicted bandgap in the range of 3.32 eV to 3.52 eV for the bulk and ultra-thin GaN, respectively. The obtained electronic band structure and density of states (DOS) of an individual unit cell of printed 2D GaN are shown in Fig. 3a. Vienna Ab initio Simulation Package (VASP, version: 5.4.4) combined with the projector augmented wave (PAW) approach were utilized to accomplish the first-principles computations24–26. The Perdew-Burke-Ernzerhof (PBE) functional integrated with the DFT-D3 correction was employed to treat the exchange-functional. The plane wave’s cut-off energy was adjusted at 520 eV27. The Brillouin zone integration was accomplished with 15*15*6 Monkhorts-Pack point sampling to optimize the bulk GaN. The self-consistent computations can give a 10-4 eV convergence energy threshold. The optimal values of the equilibrium geometries and lattice constancies were obtained with maximum stress on all atoms in 0.01 eV/Å. For the GaN (110) surface structure, we use the 7*7*1 K-points for structural optimization and self-consistent calculations. Because the PBE functional will underestimate the band gap of the semiconductor, we also use the hybrid functional method (HSE06)28 to calculate the band gap and DOS. Our density PBE functional analysis indicates that the GaN has a 1.51 eV direct bandgap. According to the HSE06 calculation, the printed 2D GaN has a 3.32 eV direct bandgap. From Fig. 3b, the measured band gap derived from the UV-Vis absorption is 3.3 eV, which agrees well with the value derived from HSE06 calculation. By comparison, the calculated bandgap using energy density PBE functional analysis is smaller than the measurement result. Figure 3c shows a periodic slab of GaN with a non-polar (110) surface sliced from the wurtzite bulk phase.
Application of 2D GaN in electronic devices. For experimental assessment of the electron transport features of printed 2D GaN and to investigate the potential for electronic devices, field-effect transistors (FETs) were constructed to assess the 2D GaN for electronic device applications. Figure 4a shows the structure diagram of the transistor based on printed 2D GaN. We have adopted an individual side-gate design for all of the devices fabricated in this study, and detailed description of the fabrication process will be presented in the Methods section. Figure 4b presents a scanning electron microscopy (SEM) image of the device. Ag was used as gate electrodes and source–drain metal contacts, and FET channels were patterned with a width of Wch = 1000 µm and the length of Lch = 50 µm. Electrical measurements were carried out for the printed side-gated 2D GaN FET. Figure 4c presents the transfer (drain current, Ids, with respect to the gate voltage, Vgs) features of a representative 2D GaN FET. Figure 4d presents the Ids with respect to the drain–source voltage (Vds) at various values of Vgs applied to the device. As presented in Fig. 4c and d, the printed 2D GaN FET devices’ p-type switching feature with an on/off ratio is more than 105. The sub-threshold swing (SS) for the FET was 98 mV per decade, near the desired action. The average value of the room-temperature field-effect mobility (µ) was obtained as 53.1 cm2 V−1 s−1, with a mobility of 57 cm2 V−1 s−1 for the device with the best performance (The FET mobility computations are presented in the Supplementary Note 3). A statistical analysis of the performance of many FET devices on fabricated using the here reported wafer-scale printing process was conducted. Electrical features of thirty printed GaN FETs were evaluated. The mean log ON/OFF current ratio, mobility, and SS were 5.31 ± 0.52, 53.1 ± 4.5 cm2 V−1 s−1, and 97.6 ± 2.42 mV dec−1 (Fig. 4e-g), respectively. A desired was seen between various devices, considering that the mentioned devices were constructed in an academic laboratory. This uniformity yields confidence in employing the mentioned devices and methods in numerous applications like integrated circuits and active-matrix back-planes for displays. Moreover, the 2D GaN-based FETs were significantly stable, having a high cyclability with a stable on/off ratio and stable on currents for more than 100 switching cycles under ambient conditions (Supplementary Fig. 4).
Notably, the observed field-effect mobility is greater compared to that of traditional high-efficiency broad-bandgap GaN-based devices14, 29. For further broader comparison, Fig. 5 plots SS versus mobility for p-type and n-type FETs reported in the literature as well as the device from this study, the value of SS is comparable to that appeared in some of the best p-type and n-type oxide semiconductors like indium oxide, this highlights the fact that our approach offers a high current mobility while maintaining a small SS, which is mainly attributed to the high-quality electron-level GaN semiconductor. Besides, no detectable performance degradation was observed in the devices while working under ambient situations without encapsulation. This highlights that the printed GaN is of exceptional quality while indicating that additional enhancements can be achieved via the enhanced device construction. Future studies should concentrate on constructing devices that can integrate separately addressable transistors into more complicated circuits. Potential approaches can involve using vdW heterostructures with dielectrics like hexagonal boron nitride or Ga2O3 combined with top gates30, 31. The mentioned device configurations can be employed to determine essential parameters like the pinch-off voltage that can be informative for incorporating future devices into functional circuits.