I. INTRODUCTION

Solid-state ultraviolet (UV) detectors have received much attention recently due to their miniature size, light weight, and low production cost. They have been developed to replace bulky, fragile photomultiplier tubes. Al_xGa_1-xN semiconductor alloys would be suitable materials for fabricating solid-state UV detectors for two reasons: (1) the energy bandgap can be tuned between 3.3 and 6.2 eV by adjusting the Al mole fraction and (2) this large energy bandgap allows low dark current. The energy bandgap and its cut-off wavelength as a function of the Al mole fraction are shown in Fig. 1, whose relationship is given in ref. ⁽¹⁾. Additional advantages include direct bandgap, high quantum efficiency, physical and chemical stability, a high breakdown field, and the ability to operate at high temperature ⁽²⁾. AlGaN photodiodes are used in solar-blind UV detection, including flame detection, chemical and biological agent detection, partial discharge detection, ozone-hole detection, secure short-range communication, and missile plume detection ^(3-6). Several research groups have demonstrated solar-blind photodetectors based on AlGaN material systems ^(7-12).

Fig. 1. Energy bandgap and cut-off wavelength of Al_xGa_1-xN alloys as functions of Al mole fraction.

In this study, we designed and fabricated AlGaN PIN photodiodes that can detect UV emissions ranging from 230 nm to 270 nm. The photodiode was designed to eliminate the need for an external optical filter.

II. EXPERIMENTS AND DISCUSSION

The epitaxial structure of the AlGaN photodiode consisted of an 8 nm highly-doped p-GaN layer (N_A = 1 × 10²⁰ cm^-3), a 20 nm p-AlGaN grading layer (N_A = 1 × 10¹⁸ cm^-3), a 200 nm p-Al_0.6Ga_0.4N layer (N_A = 8 × 10¹⁶ cm^-3), a 100 nm i-Al_0.46Ga_0.54N layer (N_D = 1 × 10¹⁶ cm^-3), a 1000 nm n-Al_0.6Ga_0.4N layer (N_D = 8 × 10¹⁸ cm^-3), a 1000 nm interlayer, and a 3000 nm AlN layer on top of a 1 mm sapphire substrate. The epitaxial structure was designed for selective detection of the wavelength range of interest. The light absorption layer was i-Al_0.46Ga_0.54N layer, whereas the upper and lower parts had wider energy bandgaps to be transparent enough for the wavelength of interest.

The photodiode fabrication was carried out as follows. A mesa isolation was defined by reactive ion etching (RIE) with a BCl₃Cl₂ gas mixture. The n-type AlGaN ohmic contact was formed by Ti/Al/Ni/Au metallization followed by rapid thermal annealing at 820 °C for 30 sec in a nitrogen ambient, whereas the top p-type GaN ohmic contact was formed by Ni/Au metallization that was annealed at 500 °C for 1 min in a nitrogen ambient. A 270 nm plasma-enhanced chemical vapor deposition SiO₂ film was deposited at 350 °C for surface passivation. Finally, a Ti/Au metal stack was evaporated for pad electrodes. The cross-sectional schematic of the AlGaN PIN photodiode is shown in Fig. 2(a), and the corresponding energy band diagram is shown in Fig. 2(b).

Fig. 2. (a) Cross-sectional schematic, (b) the corresponding energy band diagram.

The spectral photoresponsivity characteristics of a fabricated AlGaN photodiode with different bias conditions are shown in Fig. 3. The incident light was illuminated from the top front side. A peak photoresponsivity of 125 mA/W and external quantum efficiency of 60% were achieved at 245 nm under zero bias condition, which is the highest quantum efficiency achieved from AlGaN photodiodes to the best of our knowledge ^(13-16). The spectral characteristics exhibited a selective photoresponsivity band between 230 nm and 270 nm without an external optical filter, which is suitable for partial discharge detection in power plants or power transmission lines ⁽¹⁷⁾. When a reverse-bias voltage was applied to the photodiode, the photoresponsivity was enhanced noticeably by the drift component of photogenerated carriers. The maximum photoresponsivity was 170 mA/mm with an external quantum efficiency of 80%.

Fig. 3. (a) Photoresponsivity, (b) external quantum efficiency characteristics of a fabricated AlGaN photodiode as a function of reverse bias voltage. The inset is the microscopic image of a fabricated photodiode with a diameter of 250 μm.

The dark current and photocurrent characteristics of a fabricated photodiode as a function of the reverse bias voltage, where the photocurrent was measured at a wavelength of 245 nm, are shown in Fig. 4. Under zero bias condition, the dark current was 1.3 × 10^-8 A/cm², allowing very weak UV emission detection with zero standby power. We suggest that the gradual increase in the dark current as increasing the reverse bias voltage were attributed to the dislocation-induced leakage component. Further optimization of the epitaxial growth technique is required to suppress the leakage current.

Fig. 4. Dark current and photocurrent at 245 nm as a function of reverse bias voltage.

III. CONCLUSIONS

We developed a filter-free AlGaN photodiode for partial discharge detection. A wavelength between 230 nm and 270 nm was selectively detected by a proposed AlGaN photodiode. A peak responsivity of approximately 125 mA/W was achieved at 245 nm with a low dark current density of 1.3 × 10^-8 A/cm² at zero bias condition; these are state-of-the-art characteristics. The proposed photodiode does not need an external filter and can offer zero standby power.