1. Introduction

In the last decade, GaN-based laser diodes have garnered significant attention owing to their potential applications in various fields such as laser displays, laser lighting, laser storage, and visible light communication^[1−4]. These applications require higher output power, enhanced thermal stability, and increased modulation bandwidth^{[5, 6]}. One of the critical challenges lies in achieving a balance between higher injection current density and lower series resistance. Effective solutions are crucial for addressing this challenge. During the fabrication of GaN laser diodes, electrode deposition and annealing play crucial roles in achieving the desired low series resistance and optimal wall-plugging efficiency^[7−11].

In Ⅲ−nitride materials, there are substantial Schottky barriers between GaN and its corresponding metal electrode. This challenge becomes particularly pronounced when a positive bias is applied to the p−n junction, leading to the negative biasing of two Schottky barriers. To enable efficient current injection from the metal electrode, it becomes essential to establish an Ohmic contact at the interface. Researchers often chose Ti/Al as the n-electrode. Such material combination has demonstrated Ohmic contacts with n-GaN^[12−14], resulting in specific contact resistance levels as low as 10⁻⁶ to 10⁻⁸ Ω·cm².

However, research about the metallization process with p-GaN has yet to achieve a satisfying contact resistance value. Studies have showcased that high work function materials like Ni, Ti, Pd, among others, can effectively help to reduce the Schottky barrier height at the interface of metal with p-GaN. Some researchers put up with electrode schemes such as Ni/Au^[15−18], Pt/Au^[19], Cr/Au^[20], Pd/Au^[20−22], Pt/Ni/Au^[23], Ti/Pt/Au^[24], etc. Another method to reduce the Schottky barrier height is to change the composition of the contact layer^[25]. In this letter, we proposed a distinct method to reduce the specific contact resistance of the p electrode by replacing the GaN contact layer with the In_0.15Ga_0.85N contact layer. Furthermore, two commonly used electrodes Pd/Au and Ni/Au were chosen as the electrode to interpret the influence of containment of In on the contact resistance.

2. Experimental procedure

The p-GaN epitaxial layer was grown on a sapphire substrate using metal organic chemical vapor deposition (MOCVD) technique. Samples are divided into group Ⅰ and group Ⅱ. The layer structure of both groups, as depicted in Fig. 1(a), comprises a 4 μm unintentionally doped GaN buffer layer at the bottom, followed by a 20 nm AlGaN layer with a doping concentration of 1 × 10²⁰ cm⁻³, a 500 nm AlGaN/GaN superlattice with a doping concentration of 1 × 10¹⁹ cm⁻³, and a 25 nm p-doping contact layer with a doping concentration of 5 × 10²⁰ cm⁻³. This epitaxial structure closely resembles the epitaxial layer used in the laser diodes. Differences in these two groups lie in that the contact layer of group Ⅰ is GaN and the contact layer of group Ⅱ is In_0.15Ga_0.85N. To ensure the quality of the epitaxial material, the In fraction is set at 0.15. To remove the surface oxide layer which hinders the carrier transportation from metal to semiconductor^[23], the surface of the p-type GaN/In_0.15Ga_0.85N was treated with peroxymonosulfuric acid for 10 min, hydrofluoric (1% concentration) for 30 s and hydrochloric acid for 10 min before being rinsed in deionized water. The average sheet resistance of the sample with GaN contact layer and In_0.15Ga_0.85N contact layer by the non-contact Hall test are 48.90 and 49.13 Ω/□.

For sample group Ⅰ andⅡ, we fabricated the different metal stacks summarized in Table 1. The Pd thickness is 10, 20, and 30 nm for stack 1, 2, and 3, respectively. The thickness of Ni layers is 10, 20, and 30 nm for stack 4, 5, and 6, respectively. The thickness of Au is kept at 30 nm.

To assess the specific contact resistance, the circular transmission line method (c-TLM) pattern with an inner radius of 100 μm was defined using a negative photoresist. Spacing between the metal electrodes varied from 5 to 50 μm. A schematic of the c-TLM pattern is illustrated in Fig. 1(b). The Pd/Au and Ni/Au metal electrodes were deposited using the MSP-300B automatic magnetron sputtering coater, after being masked with a photoresist. The lift-off process was finalized by subjecting the samples to ultrasonic cleaning in acetone. Fig. 1(c) shows the scanning electron microscope (SEM) imagine of the Pd/Au = 10/30 nm electrode. Insert of the Fig. 1(c) is the SEM imagine of the etched cross section.

After the deposition, the samples were rapidly thermal annealed by AccuThermo AW610 rapid thermal processor at a temperature of 600 °C for 2 min under N₂ : O₂ = 4 : 1 or only N₂ atmosphere to form Ohmic contact based on past experimental experience.

3. Results and discussion

We measured the current−voltage (I−V) characteristic curves by using a Keithley 2450 source measurement unit. The total resistance was obtained through the fitting of the I−V curves using the least square method.

Fig. 2(a) presents the I−V characteristics of the sample with In_0.15Ga_0.85N contact layer using the Pd/Au = 10/30 nm electrode annealing in N₂ : O₂ = 4 : 1. Fitting of the I−V curve helps to define the total resistance. Total resistance varied with spacing between the metal electrodes is shown in the insert of Fig. 2(a). The relationship between the total resistance with ln(R/r) is plotted in the inset, where ln means the logarithm, R is the outer circle radius and r is the inner circle radius, as labeled in Fig. 1(b). Series resistance increased along with the increase of distance between two electrodes and can be well fitted into a linear relationship. Fig. 2(b) shows the I−V characteristics of the sample with GaN contact layer using the Pd/Au = 10/30 nm electrode annealing in N₂ : O₂ = 4 : 1. Total resistance versus the gap distance is shown in the insert of Fig. 2(b).

In the c-TLM, the collective resistance of the circuit can be represented by Eq. (1)^{[26, 27]}. ${R}_{\mathrm{t}}$ is the total circuit resistance; ${R}_{\mathrm{s}\mathrm{h}}$ is the sheet resistance of the sample; $R$ is the radius of the outer circle; $r$ is the radius of the inner circle; and ${L}_{\mathrm{t}}$ is the transfer distance. The transfer distance refers to the distance between the points where the current is maximum and where it decreases to 1/e of this maximum. This can be mathematically expressed by Eq. (2).

$${R}_{\mathrm{t}}=\frac{{R}_{\mathrm{s}\mathrm{h}}}{2\pi }\times \left[\mathrm{ln}\left(\frac{R}{r}\right)+{L}_{\mathrm{t}}\left(\frac{1}{R}+\frac{1}{r}\right)\right].$$

(1)

$${L}_{\mathrm{t}}=\sqrt{\frac{{\rho }_{\mathrm{c}}}{{R}_{\text{sh}}}}.$$

(2)

$${\rho }_{\mathrm{c}}=\frac{\pi {r}^{2}{{R}_{0}}^{2}}{2K}.$$

(3)

The specific contact resistance (${\rho }_{\mathrm{c}}$) can be derived using Eq. (3) by treating $\mathrm{ln}\left(\dfrac{R}{r}\right)$ as an independent variable. The slope of the curve ($K$) was determined as $\dfrac{{R}_{\mathrm{s}\mathrm{h}}}{2\pi }$. Notably, at $\mathrm{ln}\left(\dfrac{R}{r}\right)=0$, the total resistance (${R}_{\mathrm{t}}$) equals ${R}_{0}$. According to the deduction of the above results, when using the same electrode of Pd/Au = 10/30 nm, the specific contact resistance of group Ⅰ and group Ⅱ are 7.61 × 10⁻³ and 2.57 × 10⁻⁵ Ω·cm², respectively. As a conclusion, replacing the GaN contact layer with In_0.15Ga_0.85N can help to reduce the specific contact resistance effectively.

To explore the mechanism for the reduced contact resistance of the In_0.15Ga_0.85N contact layer compared with the GaN contact layer, we simulate the band structure of the contact material for the interface between the metal electrode and the semiconductor. Schematic of the band diagrams are shown in the Fig. 3. The barrier height (${\Phi }_{\mathrm{B}\mathrm{P}}$) for the transport of holes at the interface of a Pd/p-type semiconductor can be determined by the In composition of the contact layer.

As shown in Fig. 4, the barrier height varies with the In composition of the contact layer. Valence band level increases along with the increase of the In fraction^[28]. Work function of Pd is fixed at 5.12 eV from the vacuum energy. An increase of the In composition leads to the reduction of the barrier height. Holes could penetrate the interface with lower barrier height more easily which demonstrated to be with lower specific contact resistance.

To develop an optimal metal electrode combination on the In_0.15Ga_0.85N contact layer, the Pd/Au and Ni/Au electrodes which are most commonly used in the formation of Ohmic contact with p-GaN are investigated. The thickness of Pd and Ni was changed from 10 to 30 nm.

As shown in Fig. 5(a), there is a dramatic increase in specific contact resistance as the Pd thickness increases to 20 nm. As demonstrated in the related studies, a plot of current vs. voltage should produce a straight line if tunneling is dominating condition mechanism^[20]. An increase of the Pd thickness to 20 nm leads to a non-linear relationship for the I−V curve. The tunneling effect could have arisen from the acceptor-type defects produced by the reaction of V_N with the neighbor host Ga atom. Fermi level position shifts to the energy level of acceptor defects, Ga_N and V_Ga, where Ga_N is the antisite defect and V_Ga is the Ga vacancy because their energy levels are located at the vicinity of the valence band^{[29, 30]}. An increase in the Pd thickness may block the escape of N atoms along with suppression of the acceptor-type defect. Thus, the tested specific contact resistance is much higher than the sample with 10 nm Pd. Adopting the Pd/Au = 10/30 nm facilitates to achieve the lowest specific contact resistance of 2.57 × 10⁻⁵ Ω·cm².

As shown in Fig. 5(b), there is an increase in the specific contact resistance as the Ni thickness changes from 20 to 30 nm. Since the crystalline NiO and the amorphous Ni−Ga−O phases play an important part in low resistance Ohmic contact to p-GaN^{[15, 31, 32]}, a 30 nm Ni layer is too thick to be fully oxidized in the 2 min annealing time.

As to the influence of the annealing atmosphere, the results under different conditions for Pd/Au behave little difference but the specific contact resistance of Ni/Au for annealing in N₂ is much higher than in N₂ : O₂ = 4 : 1.

To find out the influence of the annealing atmosphere on the specific contact resistance, we performed the auger electron spectroscopy (AES) measurement.

As shown in Figs. 6(a) and 6(b), the atomic distribution is almost the same in different annealing atmospheres for Pd/Au. As to the Ni/Au electrode, there is large-scale diffusion of Ni through the Au capping layer to the surface of the contact where it oxidized annealing in N₂ : O₂ = 4 : 1 atmosphere. However, this reaction mechanism might play an insignificant role when annealing in the N₂ atmosphere.

4. Conclusions

In this study, we investigate the effects of substituting In_0.15Ga_0.85N for GaN as the contact layer on the specific contact resistance of the interface between a p-type semiconductor with a metal electrode. Introduction of In composition leads to the reduction of the barrier height. Furthermore, adopting the Pd/Au = 10/30 nm as the metal electrode, we achieved the specific contact resistance of 2.57 × 10⁻⁵ Ω·cm², which is a comparable value with those from the n-GaN electrode. Effects of the doping level on the depletion width and barrier height are also investigated by simulation. Finally, the Pd/Au and Ni/Au electrodes which are most commonly used in the formation of Ohmic contact with p-GaN are investigated. Variation of the specific contact resistance with metal thickness is studied. AES results help to explain the mechanism of the variation of specific contact resistance under different annealing atmospheres.

Acknowledgement

This research was partially funded by the Natural Science Foundation of China Project, grant number 62274042; Natural Science Foundation of Shanghai, grant number 21ZR1406200; the Key Research and Development Program of Jiangsu Province, grant number BE2021008-5.