University of California Santa Barbara (UCSB) and National Taiwan University (NTU) have been optimizing ammonia-based molecular beam epitaxy (MBE) for gallium nitride (GaN) growth on a range of substrates [Erin C. H. Kyle et al, J. Appl. Phys., vol115, p193702, 2014].
The researchers claim their highest single-die electron mobility of 1265cm2/Vs at 296K ('room temperature', RT) for films grown on low-threading-dislocation-density (TDD) free-standing (FS) GaN templates as "the highest RT bulk GaN electron mobility to date". To back up the claim, the paper quotes a number of results from other groups (Table 1).
Table 1: Room-temperature mobility of GaN grown by a variety of methods by different groups.
Method Mobility (cm2/V-s) Year HVPE 1245 2001 MOCVD 1005 2006 Ammonothermal 265 2007 N-rich PAMBE 1150 2007 Ga-rich PAMBE 1191 2000 NH3-MBE 560 1999 UCSB/NTU 1265 2014
The researchers used a Veeco Gen 930 molecular beam epitaxy system with an unheated showerhead injector for the delivery of purified ammonia (NH3). The gallium, silicon and magnesium came from standard effusion cells.
The templates were isolated from the active region of the bulk GaN by a structure consisting of an intrinsic layer sandwiched between lightly and heavily Mg-doped p-type GaN regions (Figure 1). The isolation ensured that negligible current flowed through the template or re-growth interface. Heavily Mg-doped GaN tends to result in rough surfaces. The undoped and lightly doped top layers of the isolation gave a smooth surface for further growth.
The bulk of the structure consisted of lightly Si-doped n-type GaN with a final layer of heavily doped material "to facilitate the formation of high-quality low-resistance ohmic contacts".
Figure 2: Average bulk electron mobility (solid blue diamonds) and average carrier concentration (red open squares) as function of active region growth temperature.
The growth was optimized with respect to temperature (760-880°C) and Si-doping concentration (~3x1016 ? ~2x1020/cm3). Lumilog provided the semi-insulating iron-doped GaN:Fe on sapphire templates. The ammonia flow rate during optimization was 200 standard cubic centimeters (SCCM), giving a growth rate of 7.4nm/minute. The highest mobility of more than 700cm2/V-s occurred in 820°C growth (Figure 2). The optimum silicon doping was found to be ~3x1016/cm3 (Figure 3). Mobility decreases with increased doping, while some doping is needed to make an ohmic contact with metal electrodes.
Figure 3: Effect of carrier concentration on electron mobility for GaN grown with optimized growth conditions.
Further experiments (Table 2) involved growth on a variety of templates where the ammonia flow rate was 200 or 1000 (SCCM) with the aim of quantifying the effects of TDDs. Naturally, the highest-mobility results came from using free-standing GaN templates with low TDDs. The researchers carried out a wide range of experimental and theoretical analyses to explore the impact of TDDs on the electrical performance of the GaN film.
Table 2: Single-die Hall measurements for 200SCCM and 100SCCM ammonia flow TDD series. Full-width at half maximum (FWHM) is for ω-scan x-ray diffraction from GaN (20-21) planes.
TDD-200 Series
TDD (/cm2) RT mobility (cm2/Vs) RT carrier concentration (/cm3) Highest mobility (cm2/Vs) Highest mobility temp. (K) FWHM (arcsec) ~3x107 1256 4.48x1016 2948 116 225 ~5x108 961 3.50x1016 2396 115 382 ~5x109 204 4.9x1016 343 154 739
TDD-1000 Series
TDD (/cm2) RT mobility (cm2/Vs) RT carrier concentration (/cm3) Highest mobility (cm2/Vs) Highest mobility temp. (K) FWHM (arcsec) ~2x106 1265 3.73x1016 3327 113 90 ~5x108 966 2.09x1016 2637 112 375 ~2x1010 317 1.33x1017 348 212 1370
