1 Introduction

Since recent great improvements in the luminescence efficiency of GaN-based light-emitting diodes (LEDs), their applications have ranged over wide area including display backlights, traffic lights, and solid-state lighting. Bi-layer metals like a Ni/Au layer with transparency <70 % have been extensively used as a transparent contact layer on p-GaN in LEDs [1], but its low transparency is considered to be one of the major factors limiting the efficiency of the LEDs. Several transparent conducting oxide (TCO) materials such as impurity-doped SnO2, In2O3, CdSnO3, CdIn2O4, Zn2In2O5, and In4Sn3O12 multi-component compounds have been studied as alternatives to the bi-layer-metal contacts to overcome their drawback: low transparency [2]. Among the TCOs, tin-doped indium oxide (ITO) is now being widely used due to its low resistivity (~10−4 Ω cm) and high transparency in the visible range [3]. However, ITO has inherent problems such as high cost, limited supply of indium, and thermal instabilities, thereby making its use as TCO in LEDs become difficult [4]. ZnO-based TCO is also a potential alternative to serve as a transparent contact layer in LEDs. Despite similar electrical and optical properties with ITO, ZnO-based TCO films exhibit several advantages over ITO, including better thermal conductivity, much smaller lattice mismatch (1.8 %) to GaN, superior high-temperature stability, and higher transparency due to large bandgap (3.37 eV) [5–9]. The conductivity of undoped ZnO is low, but can be enhanced by doping III-group elements such as Al, In, and Ga [10]. Most research has been performed on Al doping in ZnO, but Ga seems to be the better dopant for ZnO due to close covalent bond lengths of Ga-O and Zn–O (the bond length is 1.92, 1.97, 2.7, and 2.1 Å for Ga–O, Zn–O, Al–O, and In–O, respectively) and its lower reactivity with oxygen [11]. Several studies have been performed on the properties of Ga-doped ZnO (GZO) as TCO [7, 12, 13], but very rare on the luminescence from GaN-based LEDs with GZO contacts. In this paper, we study luminescence and I–V characteristics of p–n junction GaN LEDs with GZO contacts for various Ga concentrations in relation to the effects of Ga concentration on structural, electrical, and optical properties of GZO.

2 Experimental

LED structures with GaN p–n epilayers were grown on c-plane (0001) sapphire substrates by metal–organic chemical vapor deposition (MOCVD). From bottom to top, this LED structure consisted of a 50-nm-thick GaN nucleation layer, a 3-μm-thick unintentionally doped GaN layer, a 3-μm-thick Si-doped GaN n-cladding layer, and a 70-nm-thick Mg-doped p-GaN layer. Half region of the p-GaN layer was etched until the n-GaN layer was exposed for the deposition of GZO contacts. A mixture of ZnO/Ga2O3 powders was compressed to form 2-inch-diameter discs by varying Ga concentration (n G ) from 1 to 5 mol%, and then sintered at 600 °C in vacuum for 2 h to yield sputtering targets. The targets were then mounted in a RF magnetron sputtering system to deposit 100-nm-thick GZO films of 500 μm diameter as electrodes on both p- and n- GaN layers using a shadow mask. After the system had been evacuated to a base pressure of 3 × 10−6 Torr, the substrates were pre-sputtered at a power 75 W for 10 min. The GZO sputtering was then performed at 400 °C for 50 min at a working pressure of 10−3 Torr. Other deposition conditions are as follows: RF power: 105 W, deposition rate: 2 nm/min, and gas mixing ratio of O2/Ar: 1:8. Finally, the GZO contacts were annealed at 900 °C for 3 min in a N2 ambient to activate Ga dopants and enhance the crystal quality of GZO electrodes. GZO films were also deposited on n-type Si wafers under same preparation conditions for Hall measurements.

Hall-effect measurements of GZO films on n-type Si for various Ga concentrations were performed in an apparatus (Ecopia model HEM-2000) using the Van der Pauw method. For Ohmic contacts, an In film of 500 μm diameter was deposited on GZO films using a shadow mask in a thermal evaporator and was subsequently annealed at 200 °C for 10 min. The intensity of the B field was 0.37 T, and the current was varied from 0.1 μA to 10 mA. All Hall measurements were performed in a dark room to exclude ambient light. PL spectra were measured using the 325-nm line of a He-Cd laser as the excitation source. Emitted light was collected by a lens and analyzed using a grating monochromator and a GaAs photomultiplier tube. Standard lock-in detection techniques were used to maximize the signal-to-noise ratio. The laser beam diameter was about 0.3 mm and the power density was about 2.12 W/cm2. The current–voltage (I–V) curves of the LED devices were measured at room temperature (RT) using Keithley 237 high-voltage source meter. For forward-bias conditions, positive voltages were applied to the GZO contacts on the p-GaN layers. The RT electroluminescence (EL) measurements were performed under forward bias using DM 500 grating monochromator.

3 Results and discussion

Figure 1 shows resistivity (ρ), electron concentration (n e ), and electron mobility (μ) of GZO films as functions of n G , obtained from the Hall-effect measurements. All GZO films exhibited n-type characteristics. The ρ shows decreasing behaviors with increasing n G and reaches a minimum value of 7.4 × 10−4 Ω cm at n G  = 2 mol%, but above n G  = 2 mol%, it monotonically increases. In a totally opposite manner, the ne and μ show maximum values of 3 × 1020 cm−3 and 28 cm2/Vs, respectively at n G  = 2 mol%. These results indicate best electrical properties of the GZO films at n G  = 2 mol%. Figure 2 shows n G -dependent PL spectra of GZO films, peaked at ~3.3 eV, resulting from near-band-edge (NBE) emissions [14]. No PL bands related to defects such as oxygen vacancies or Zn interstitial [15] were observed, meaning good-quality GZO films. The PL shows a maximum intensity at n G  = 2 mol%, as shown in the inset of Fig. 2, consistent with the Hall-effect data. Based on these electrical and optical results, it is expected that the GZO films can act as best metal contacts of LEDs at n G  = 2 mol%.

Fig. 1
figure 1

Resistivity, electron concentration, and electron mobility of GZO films as functions of Ga concentration

Full size image
Fig. 2
figure 2

PL spectra of GZO films for various Ga concentrations. The inset summarizes the intensity of the NBE PL as a function of Ga concentration

Full size image

Figure 3a shows XRD spectra of GZO films for various doping concentrations. The XRD band at 34.5°, corresponding to ZnO (002) plane [16], is dominantly observed in all samples. This peak shows a maximum intensity at n G  = 2 mol%, meaning best crystal quality, as shown in Fig. 3b. This possibly suggests that Ga atoms are almost completely substituted for Zn sites at n G  = 2 mol%, resulting in the formation of large crystallites, thereby reducing electron scattering at the boundaries [17, 18], consistent with the best electrical and optical properties at n G  = 2 mol%, as shown in Figs. 1 and 2. At n G  ≥ 3 mol%, another XRD peak, known to come from ZnGa2O4 (311) plane [19], appears at ~35.3°, indicating that the solubility of Ga in ZnO is <3 mol%. This secondary phase could distort the lattice or produce point/line defects responsible for carrier capturing/scattering, resulting in degradation of the electrical properties/reduction of the PL intensity at n G  ≥ 3 mol%, as shown in Figs. 1 and 2. In Fig. 3a, the (002) XRD peak moves to lower 2θ values as n G increases, possibly resulting from increased distortion of ZnO lattice by the formation of Ga antisites or interstitials at n G beyond the solubility.

Fig. 3
figure 3

a XRD patterns of GZO films for various Ga concentrations. b XRD intensity as a function of Ga concentration

Full size image

Figure 4 shows I–V characteristics of p–n junction GaN LEDs with GZO contacts. Positive voltages were applied to the GZO layer on p-GaN under forward bias. For n G  = 1–4 mol%, the I–V curves show diode characteristics. Almost no current flowed in the devices with undoped ZnO layer and no diode characteristics were observed in the devices with n G  = 5 mol%. As shown in the inset, the device shows smallest leakage current at n G  = 2 mol% because the defects such as antisites and interstitials were minimized by the effective substitution of Ga atoms for Zn sites at n G  = 2 mol%, as explained above. The unnecessary defects could produce capture sites or leakage paths, which explains why the leakage currents increase at n G  ≥ 3 mol% and the diode characteristics are almost lost at n G  = 5 mol%. The onset voltages are estimated to be 3.4, 3.0, 3.1, and 4.9 V for n G  = 1, 2, 3, and 4 mol%, respectively. These values are comparable to those of the conventional GaN LEDs except n G  = 4 mol%.

Fig. 4
figure 4

I–V characteristics of the LED devices. The inset shows reverse-bias leakage current as a function of Ga concentration

Full size image

Figure 5a shows EL spectra of p–n junction GaN LEDs with 2 mol% GZO contacts. Two dominant EL peaks of ultraviolet (UV) and yellow emissions are observed at ~376 and ~560 nm, respectively. Both EL peaks can be attributed to radiative recombinations between N-vacancy-related shallow donor located at ~100 meV below the conduction band and the valence band edge and between defects at the p-GaN/n-GaN boundaries and in the GaN lattice, respectively [20]. Despite the increase of the peak intensities at larger injection currents, negligible energy shifts in both peaks are observed, indicating no device fatigue by the high-current-induced heat. The inset shows a photo of luminescence from the LED at a current of 10 mA, taken by a camera (SONY cyber-shot dsc-T70) under dark ambient. White light is observed around the GZO contacts and is thought to be very strong because the electrical tip is seen even in dark by the naked eye. The yellow emissions can be attributed to the defects, as explained above, judging from their coming from the p-GaN/n-GaN boundaries. Fig. 5b shows the UV EL intensity as a function of injection current for various Ga concentrations. The EL intensities increase with increasing current irrespective of n G . At n G  = 2 mol%, the EL shows remarkably large intensities for all currents, and at n G  = 5 mol%, no EL is observed for the currents <3 mA, consistent with the results of PL, I–V, and XRD.

Fig. 5
figure 5

a RT EL spectra of 2 mol% GZO/p-GaN/n-GaN/2 mol% GZO LED structures as a function of injection current. The inset shows EL image at an injection current of 10 mA. b UV EL peak intensities as functions of injection current for various Ga concentrations

Full size image

Figure 6a and b shows band diagrams of GZO/p-GaN/n-GaN/GZO LED structures under unbiased and forward-bias conditions, respectively, based on Anderson model [21]. The bandgaps (Eg ZnO and Eg GaN) of ZnO and GaN are 3.37 eV and 3.4 eV with electron affinities (χZnO and χGaN) of 4.35 eV and 4.2 eV, respectively, and therefore, the discontinuities of the conduction band and valence band, ΔEC (= χZnO − χGaN) and ΔEV (= Eg ZnO − Eg GaN + ΔEC), are deduced to be 0.15 and 0.12 eV, respectively [22]. Since the Fermi level should be constant in all layers under zero bias, the energy band of p-GaN is bent like an arch, as shown in Fig. 6a. Once a positive voltage is applied to the GZO contact on p-GaN, the GZO/p-GaN junction is reverse-biased, whilst the p-GaN/n-GaN junction is forward-biased, resulting in a voltage drop at the GZO/p-GaN junction, thereby inducing downward band bending in the direction from p-GaN to GZO, as shown in Fig. 6b. By this band bending, the excess electrons injected from the negative GZO contact flow to the GZO/p-GaN and recombine with the excess holes injected from the positive GZO contact, thereby producing UV EL.

Fig. 6
figure 6

a Zero- and b forward-biased energy-band diagrams of the LED device. All numbers in a are indicated in unit of eV

Full size image

4 Conclusion

MOCVD and RF sputtering were employed to fabricate the GZO/p-GaN/n-GaN/GZO LED structures, where the GZO films were used as electrical contacts. The GZO films showed best electrical, optical, and structural properties at n G  = 2 mol%, as confirmed by Hall effect, PL, and XRD. While positive voltages were applied to the GZO contacts on the p-GaN layers, all devices except n G  = 5 mol% showed rectifying I–V characteristics, and the leakage current was minimized at n G  = 2 mol%. Two major EL emissions observed in the UV (~376 nm) and yellow ranges (~560 nm) were attributed to NBE- and defect-related radiative transitions, respectively. The UV EL showed highest intensities over full range of injection current at n G  = 2 mol%, consistent with the results of Hall effect, PL, I–V, and XRD. The EL mechanisms were explained based on the energy band diagrams of the GZO/p-GaN/n-GaN/GZO structures.