Effect of Inductively Coupled Plasma Etching Parameters on n-Al0.5Ga0.5N Ohmic Contact

High-Al-content n-AlGaN ohmic contact is very important for deep ultraviolet optoelectrical devices. However, it often suffers from the etching damages formed in inductively coupled plasma (ICP) etching. In this paper, the effects of ICP etching parameters on n-Al0.5Ga0.5N ohmic contact, including RF power, ICP power, and etching gas, were systematically investigated and analyzed by X-ray photoelectron spectroscopy and circular transmission line model. Finally, n-Al0.5Ga0.5N ohmic contact was achieved with a low specific contact resistivity of 8.7×10-4 Ω·cm2, and AlGaN-based UVC light-emitting diodes (LEDs) showed a low operation voltage of only 5.6 V at the injection current density of 16 A/cm2.

Effect of Inductively Coupled Plasma Etching Parameters on n-Al 0.5 Ga 0.5 N Ohmic Contact Shanshan Yang , Meixin Feng , Yuzhen Liu, Wenjun Xiong , Biao Deng, Yingnan Huang, Chuanjie Li , Qiming Xu, Yanwei Shen, Qian Sun , and Hui Yang Abstract-High-Al-content n-AlGaN ohmic contact is very important for deep ultraviolet optoelectrical devices.However, it often suffers from the etching damages formed in inductively coupled plasma (ICP) etching.In this paper, the effects of ICP etching parameters on n-Al 0.5 Ga 0.5 N ohmic contact, including RF power, ICP power, and etching gas, were systematically investigated and analyzed by X-ray photoelectron spectroscopy and circular transmission line model.Finally, n-Al 0.5 Ga 0.5 N ohmic contact was achieved with a low specific contact resistivity of 8.7×10 -4 Ω•cm 2 , and AlGaN-based UVC light-emitting diodes (LEDs) showed a low operation voltage of only 5.6 V at the injection current density of 16 A/cm 2 .Index Terms-N-AlGaN, ohmic contact, ICP etching, XPS.

I. INTRODUCTION
A LGAN materials are very suitable for the fabrication of ultraviolet (UV) optoelectronic devices, such as UV LED, which can be widely used in sterilization, water purification, UV curing, specialty lighting, bio-phototherapy etc. [1], [2], [3], [4].However, in UV optoelectrical devices, with the increase of Al molar fraction in AlGaN material, the energy bandgap increases, which makes that low-resistivity ohmic contact is more and more difficult to obtain [5], [6], [7].On the other hand, to fabricate UV LEDs or laser diodes, ICP dry etching is often utilized to expose the buried n-AlGaN layer for the fabrication of ohmic contact [4], [8], [9], [10], [11], [12].Therefore, it is very crucial to study the effect of etching parameters on n-AlGaN ohmic contact to achieve a low specific contact resistivity for the fabrication of high-performance UV optoelectronic devices [13], [14], [15], [16].
During the dry etching process, the distribution and state of surface elements in AlGaN material changes, leading to the generation of defects, such as vacancies and dangling bonds [17], [18], [19].For n-type GaN, the nitrogen vacancy induced by dry etching often acts as a donor, and then the Fermi energy level in n-GaN moves upward after plasma treatment, which facilitates the formation of n-type ohmic contact [20], [21], [22], [23].However, this situation is different for n-type AlGaN with high Al content, where the etching damage tends to become a defect with deep energy level, making ohmic contact very difficult and hence affecting the electrical characteristic of the device [24], [25], [26], [27].Previous studies reported that plasma pretreatment of n-AlGaN before metal contact was used to change the material quality and morphology of the etched surface.However, few efforts have been made to systematically study the effects of plasma etching parameters on the ohmic contact of n-AlGaN with high aluminum components [8], [28], [29], [30].
This paper methodically investigates the impact of ICP etching parameters on high-Al-content n-AlGaN metal contact.CTLM and XPS were used to characterize the specific contact resistivity, Ga 3d core level and surface stoichiometry of n-AlGaN sample.As a result, an optimal ICP etching condition was given and showed a low-specific contact resistivity for n-Al 0.5 Ga 0.5 N. Furthermore, this optimal ICP etching condition was utilized in practical AlGaN-based UVC LEDs.

II. EXPERIMENTAL METHODS
Fig. 1 illustrates a schematic diagram of the n-Al 0.5 Ga 0.5 N samples used in this study.The epitaxial layers were grown by using metal-organic chemical vapor deposition (MOCVD) method.and consisted of an AlN/AlGaN multilayer buffer, a 2-μm-thick unintentionally doped AlGaN layer, and a 2-μm-thick Si-doped n-Al 0.5 Ga 0.5 N layer with a Si doping concentration of 1×10 19 cm -3 .Hall measurement showed that

TABLE I VARIOUS ICP ETCHING CONDITIONS FOR EIGHT N-ALGAN SAMPLES
the effective electron concentration of n-Al 0.5 Ga 0.5 N layer was 1×10 19 cm -3 .The manufactured n-AlGaN wafer was cut into eight individual samples to study the effect of ICP etching parameters on n-Al 0.5 Ga 0.5 N ohmic contact as shown in Table I.
Prior to metal deposition, the samples were treated by using the ICP dry etching method in the Oxford Plasmalab system at 20 °C for 4 min with a chamber pressure of 10 mTorr, and the detailed ICP etching parameters were shown in Table I.After plasma treatment, all samples were immersed in diluted HCl solution (HCl: H 2 O = 1:2) for 10 min and then rinsed with deionized water for 5 min.After that, the circular transmission line model (CTLM) patterns were transferred to the sample surface by photolithography, followed by the sputtering of Cr/Ti/Al/Ti/Ni/Au (40/30/100/70/60/80 nm) contact metal stack by magnetron sputtering as soon as possible (at intervals of less than 5 min) and the lift-off process.After rapid thermal annealing in N 2 for 2 min at a temperature of 950 °C, the samples were subjected to current-voltage (I-V) measurements by using a four-point probe technique.
CTLM patterns were used to measure the specific contact resistivity of the samples.For the CTLM patterns, the inner radius (r) was 150 μm, and the spacing (R) between the inner and outer radii varied from 10 to 80 μm.I-V tests were performed by using a KEITHLEY-2400 sourcemeter analyzer, and surface stoichiometry ratio of the dry etched samples was quantitatively characterized by using XPS with an Al Kα X-ray source energy of 1486.6 eV (ULVAC-PHI 5000 Versaprobe II).The peak energy positions of the XPS spectra were calibrated based on the position of the C1s peaks with the binding energy of 284.8 eV, and the emission angle was set to 45°.The XPS spectra were obtained by using the Shirley-type background

TABLE II SPECIFIC CONTACT RESISTIVITY AND SURFACE STOICHIOMETRY OF EIGHT SAMPLES
deduction method, and fitted with a combination of Gaussian and Lorentzian line shapes.The atomic force microscopy (AFM) image was measured by Veeco Dimension 3100.

III. RESULTS AND DISCUSSION
RF power plays an important role in ICP etching.As increasing the RF power, the plasma energy is raised, the physical attack of the plasma on the etched sample surface is enhanced, and thus the etching rate is increased.Therefore, we firstly studied the impact of RF power on n-Al 0.5 Ga 0.5 N ohmic contact by comparing samples 2, 4, and 6.As shown in Fig. 2(a), sample 2 exhibits a non-ohmic contact, while samples 4 and 6 demonstrate ohmic contact behaviors.In order to reveal the underlying mechanism, XPS analysis was performed on these samples prior to contact metal deposition.As shown in Fig. 2(b), when RF power was 100 W, the Ga 3d peak energy of n-AlGaN sample was 19.44 eV.As decreasing RF power to 50 W, the Ga 3d peak energy increased to 20.18 eV, indicating that sample 4 has a higher Fermi energy level than that of sample 2 and hence lower barrier height at the sample surface, which helps the current conduction through the tunneling and forms a low-resistivity ohmic contact.It also implies that less compensation centers of the host type were introduced, which is further proved by the result of surface stoichiometric ratio obtained from XPS analysis.As shown in Table II, the nitrogen element content was increased from 45.3% to 54.3%, indicating less nitrogen vacancies, which usually function as defects with deep energy level to compensate the surface electron concentration.These results are also consistent with the specific contact resistivity (ρ с ).As compared with sample 2, sample 4 showed an ohmic contact behavior with a specific contact resistivity of 8.7×10 -4 Ω•cm 2 obtained from the CTLM model.As further reducing RF power to 20 W, the Ga 3d peak energy of n-AlGaN sample was 19.88 eV, illustrating that the Fermi energy level of sample 6 moves downward as compared with sample 4, which increases the barrier and inhibits the current conduction.It indicates more nitrogen vacancies and higher specific contact resistivity for sample 6, which is confirmed by the results shown in Table II.As compared with sample 4, the nitrogen element content of sample 6 reduced to 52.2%, and the specific contact resistivity of sample 6 increased to 9.4×10 -4 Ω•cm 2 .In ICP etching, the molecules are accelerated by an RF source to obtain energy to bombard the sample to realize etching, and the RF power directly affects the plasma density and energy to break the Ga-N bond and Al-N bond on the surface, which induces the AlGaN etching, but also forms etching damages, such as nitrogen vacancy (V N ) [31], [32].Too high or too less RF power is not suitable to form low-resistivity ohmic contact.
The etching gas play a role in ICP etching [33].Therefore, and we studied the impact of the etching gas on n-AlGaN ohmic contact by comparing samples 3 and 4. As shown in Fig. 3(a), sample 3 exhibits non-ohmic contact, while sample 4 demonstrates ohmic contact behavior, which is mainly due to the nitrogen vacancy formed by the dry etching.As shown in Table II, sample 4 shows a higher nitrogen element content as compared with sample 3, indicating less nitrogen vacancy to compensate the surface electron concentration and hence lower barrier height, which is further experimentally proved by the Ga 3d core levels.As shown in Fig. 3(b), comparing with sample 3, sample 4 shows a higher binding energy, indicating that the energy difference between the surface Fermi energy level and valence band edge is larger, and the surface barrier height is smaller, which favors the formation of ohmic contact.In ICP etching, with inductively coupled glow discharge, the etching gas mainly becomes active free radicals, substable molecules, elements, etc.These molecules interact with AlGaN surface physically or chemically.Since physical etching easily generates nitrogen vacancy due to a smaller weight of nitrogen atom as compared with Al or Ga atoms, chemical etching is much more profitable.As compared with BCl 3 , Cl 2 could contribute larger part of chemical etching, thus pure Cl 2 etching gas is more favorable to n-AlGaN ohmic contact.
ICP power is another key parameter in ICP etching, and we studied the effect of ICP etching power on n-AlGaN ohmic contact by comparing samples 6 and 7.As shown in Fig. 4(a),  both samples 6 and 7 exhibit ohmic contact behavior.However, sample 6 had larger current and smaller resistance under the same voltage, corresponding to a lower specific contact resistivity and higher nitrogen element content for sample 6 as shown in Table II.Fig. 4(b) shows the XPS spectra of Ga 3d core levels for samples 6 and 7.It can be found that the Ga 3d binding energy of sample 6 was 0.84 eV larger than that of sample 7, indicating less surface nitrogen vacancies and lower barrier height, which contributes to a lower specific contact resistivity for sample 6.In ICP etching, under the action of a high-frequency electric field, ICP power can excite the plasma, generating electrons and ions, which etch the sample surface.A high ICP power could reduce the nitrogen vacancy and be much more preferrable for n-AlGaN ohmic contact.
Based on the above analysis, it can be seen that sample 4 showed the optimal ICP etching condition and the lowest specific contact resistivity of 8.7×10 -4 Ω•cm 2 , which is even lower that the unetched n-AlGaN sample 8 as shown in Fig. 5(a) and Table II.Fig. 5(b) shows XPS spectra of Ga 3d core levels for samples 4 and 8.The Ga 3d binding energy of sample 4 was 0.76 eV higher than that of sample 8, indicating higher surface Fermi energy level and less N vacancies, which is consistent with lower nitrogen element content as shown in Table II.It illustrates that after ICP etching, the surface electron concentration was increased, and hence the Fermi energy level moved upward to the conduction band, a low-resistivity ohmic contact was formed on the high-Al-content AlGaN sample.It should be noted that in our ICP etching, the etching gases used are Cl 2 and BCl 3 without any oxygen element, and hence the oxygen element nearly has no impact on the etching process, including the etching rate.The oxygen element obtained from XPS measurements is mainly due to the surface contamination while taking out from the ICP equipment.This is shown in Table II, the contents of oxygen element for all the measured samples was about 10%, and showed no close relationship with the measured specific contact resistance.Therefore, we thought the content of oxygen element in these samples has little effect on the measured results.
In order to verify the above experimental result, we applied the optimal ICP etching condition used for sample 4 in the practical AlGaN-based UVC LED device.The ICP and RF powers were 300 and 50 W, respectively, and the etching gas was pure Cl 2 .The detailed epitaxial structure, fabrication process, and measurement method could be found in our previous work [34].Fig. 6(a) shows the AFM image of the etched sample, showing a surface roughness of 0.4 nm, which is almost equal to that of the as-grown sample.As shown in Fig. 6(d), for a UVC LED device with a dimension of 305 × 508 μm 2 , at the injection current of 20 mA, corresponding to a current density of 16 A/cm 2 , the operation voltage was as lower as 5.6 V, and the differential series resistance was only 16.6 Ω, which was much lower than that reported in literatures [2], [35].The output power was 4.8 mW, corresponding to a high wall-plug efficiency of 4.3%.

IV. CONCLUSION
In summary, we have systematically studied the effects of ICP etching parameters on n-Al 0.5 Ga 0.5 N ohmic contact by XPS and CTLM.The experimental results showed that an appropriate RF power of 50 W, pure Cl 2 etching gas, and high ICP power of 300 W could contributes to low nitrogen vacancy concentration, high surface Fermi energy level and Ga 3d core level, corresponding to a small specific contact resistivity of 8.7×10 -4 Ω•cm 2 for n-Al 0.5 Ga 0.5 N. Additionally, this optimized ICP etching condition was utilized in practical AlGaN-based UVC LED, showing a low operation voltage of only 5.6 V at the injection current density of 16 A/cm 2 .

Fig. 5 .
Fig. 5. (a) The measured resistance as a function of ln (R/r) for samples 4 and 8. (b) XPS spectra of Ga 3d core levels for samples 4 and 8.

Fig. 6 .
Fig. 6. (a).AFM image of the sample surface after ICP etching.(b) The size of 305 × 508 µm 2 UVC LED device luminescence diagram.(c) Electroluminescence spectrum of the as-fabricated AlGaN-based UVC LED.(d) Light output power, voltages and Wall plug efficiency of UVC LED at various injection currents.