1. Introduction

Lanthanoid (Ln) doped crystals, which exhibit stable narrow-bandwidth photon emission at room temperature, are expected to be used in various applications such as bioimaging, optical sensing, lighting and display, photodetectors, optoelectronics, and quantum nanophotonics [1–3]. The photon emission properties of Ln ions in crystal are temperature-insensitive since, since the energy levels in 4f-shell which are involved in the luminescence transitions are surrounded by filled 5s and 5p orbitals and thus isolated from free carriers in the host material. In addition, the isolated Ln ions have the potential to be used for single photon source (SPS), which is one of the building blocks for quantum technologies such as quantum computing, quantum communication, and quantum sensing, due to their inherent single photon emission properties [4–10]. Ceramics and oxides have been mostly used as host materials for Ln ions due to their high solubility, while nitride semiconductors are also known as one of the superior host materials because of the potential of electric control of their photon emission (electroluminescence) [11]. Studies on the optoelectronic properties of Ln-doped gallium nitride (GaN) and the development of Ln-doped GaN light-emitting diodes (LEDs) have been conducted for decades [12]. Ln doped GaN (III-N) can be used for electrically controlled SPS operating at room temperature, which is particularly advantageous for practical applications.

Ion implantation is one of the most popular methods for Ln doping of crystals. Unlike in-situ doping during crystal growth, the ion implantation is capable of deterministic placement of Ln ions with nanometer precision [13], making it the indispensable technology for quantum device applications such as SPS. However, the drawback of ion implantation is the radiation-induced defect formation, which leads to the degradation of crystal quality. It is generally believed that implanted Ln ions are stabilized as luminescence centers by substituting into cation sites in trivalent state (Ln³⁺) [11], and the post-thermal annealing is necessary to recover the radiation induced damages. It is empirically understood that thermal annealing at temperatures higher than 2/3 of the melting point is required for complete recovery of radiation induced damage [14]. The annealing of GaN is a challenging process due to the material decomposition (preferential evaporation of nitrogen from the surface) that occurs at temperatures higher than about 850 °C at 1 atm [15]. To avoid the decomposition, special methods such as cap annealing [16–18], multi-cycle annealing [19,20], and high-pressure annealing [21–24] have been applied for implantation damage recovery, although highly efficient activation of implanted Ln ions in GaN remains a challenge.

Aluminum nitride (AlN) semiconductors, one of the nitride semiconductors which has an ultra-wide bandgap (6.0 eV), high heat conductivity and high breakdown voltages, have been extensively studied as the next generation semiconductor materials for the realization of ultra-low loss power devices and deep UV LEDs [25,26]. However, the potential as a host material for Ln ions has not been well explored because of the difficulty of AlN crystal growth, and only limited literature is currently available on the optical properties of Ln-doped AlN [27–30]. Comparing the properties as a host material with GaN, AlN is superior in high temperature resistance, and decomposition from the surface can be significantly suppressed even at 1700 °C by face-to-face setup [31–33]. Therefore, the recovery of implantation damage and highly efficient activation of implanted Ln ions are expected by such high temperature annealing, although the annealing properties of Ln implanted AlN and its optical properties have not been well studied.

Here we show the room temperature photoluminescence (PL) properties of implanted Ln ions (praseodymium: Pr, europium: Eu, and neodymium: Nd) into high quality single crystal AlN epi layers on sapphire substrates and the change of optical properties with annealing temperatures up to 1700 °C. The cathodoluminescence (CL) properties as well as the PL properties under different excitation conditions are investigated. The recovery of implantation damage by thermal annealing is analyzed by X-ray diffraction (XRD) and atomic force microscopy (AFM), and the thermal diffusion of implanted Ln ions due to high temperature annealing is investigated by secondary ion mass spectroscopy (SIMS). Frome these analyses, we clarify that there is a trade-off between the formation of Ln luminescent centers associated with the recovery of crystallinity and the increase in quenching factors such as compound defect formation and aggregation.

2. Experimental

550-nm-thick AlN films with low threading dislocation densities (TDDs) and low impurity concentrations were prepared by metalorganic vapor phase epitaxy (MOVPE) on face-to-face annealed sputtered AlN templates on c-plane sapphire substrates (FFA Sp-AlN) [31,34]. We confirmed that TDDs of MOVPE-grown AlN on FFA Sp-AlN was 2 × 10⁸cm⁻² [34]. Moreover, impurity concentrations and point defect densities of the MOVPE-grown AlN on FFA Sp-AlN were sufficiently low to eliminate the effect of background impurities [32,35]. The AlN films were implanted with Pr, Eu, and Nd ions at room temperature using the ion implanter at the Takasaki Institute of Advanced Quantum Science (TIAQ), QST, and then thermally annealed under N₂ atmosphere using the face-to-face technique to remove implantation induced defects and to activate lanthanoid ions as optical centers. The implantation angle was tilted about 7 degrees from the normal angle to avoid significant channeling effect. The implantation and annealing conditions are summarized in Table 1. The implanted 100 keV and 700 keV Pr ion profiles in AlN and GaN, calculated by the Monte Carlo simulation code TRIM [36] are shown in Fig. 1. The Pr ions are implanted 36 nm deep from the surface in the case of 100 keV and 186 nm deep in the case of 700 keV. The Pr ions are implanted deeper in AlN than in GaN (6.15 g/cm³) due to its lower atomic mass density (3.26 g/cm³). A similar implantation profile is expected for 100 keV Eu and 700 keV Nd ions, since their atomic masses are close. In addition, commercially available 3 µm thick undoped GaN epilayers on sapphire substrate were implanted for comparison. The implanted GaN samples were capped with 100 nm thick SiO₂ (chemical vapor deposition) and then thermally annealed at 1200 °C for 2 min in N₂ using an infrared furnace (rapid thermal annealing: RTA). After annealing, the SiO₂ layer was removed by hydrofluoric acid treatment (HF:H₂O = 1:5, 20 min).

 

Fig. 1. (a) Pr ion profiles implanted in AlN and GaN calculated by TRIM: 100 keV-Pr in AlN (black), 100 keV-Pr in GaN (green), 700 keV-Pr in AlN (red), and 700 keV-Pr in GaN (blue). The average range and the mass densities of AlN and GaN are shown in the figure. The implantation dose is set to be 1 × 10¹⁴cm^-2. (b) Implantation induced damage (dpa) profiles at the dose of 1 × 10¹⁴cm^-2. The displacement energies are 25 eV for Al and 28 eV for N.

Download Full Size | PDF

Table 1. Ion implantation and thermal annealing conditions

View Table

The room temperature optical properties of implanted Ln ions were characterized by laser scanning confocal microscopy (microphotoluminescence measurement). The excitation laser wavelengths were 266 nm for 4f5d excitation of Pr ions and Eu ions, and 527 nm and 785 nm for resonant excitation of Pr and Nd ions. The time-resolved PL (TRPL) measurement was performed to analyze photon emission lifetimes using a Si single-photon avalanche photodiode (Si-APD) and a time-correlated single photon counting (TCSPC) system. An appropriate bandpass filter was used to selectively collect photons emitted by Ln ions. For pulsed diode lasers, a mechanical laser chopper or a modulation pulse was used. The cycle duration and time bin were 1 kHz and 51.2–102.4 ns, respectively. Cathodoluminescence (CL) measurements were also performed for Pr-implanted AlN at room temperature. An Acceleration voltage was set to 5 kV. To analyze the implantation induced damage and the recovery by thermal annealing, X-ray diffraction (XRD) 2θ-ω profiles around AlN (0002) were measured to evaluate the structural properties. The surface morphology was characterized by an atomic force microscopy (AFM). Secondary ion mass spectrometry (SIMS) was performed on Pr-implanted AlN samples to analyze the change in implanted ion profiles with thermal annealing. All SIMS experiments were performed on a ULVAC-PHI ADEPT-1010 quadrupole SIMS instrument using O₂⁺ primary ions. The O₂⁺ primary ion beams were accelerated to 5 kV. Charge correction of the sample surface was performed using an electron gun. Positive secondary ions of M⁺ produced by O₂⁺ primary beams were detected by electron multiplier. The Pr profile was quantified by analyzing Pr ion implanted samples under the same conditions.

3. Results

3.1 Optical properties of implanted Pr and Eu ions under excitation to the 4f5d band

An electron in the ground level (³H₄ level) of Pr³⁺ can be excited to the 4f5d band by a 266 nm (4.66 eV) laser and 5 keV electron irradiation [37–39]. The excited electron then rapidly relaxes to a level in the 4f shell and shows photon emission as the electron transitions to the ground level or another level in the 4f shell. Figures 2 (a) and (b) show PL and CL spectra of 100 keV Pr implanted AlN at the dose of 3 × 10¹⁴cm^-2 after thermal annealing at 1200 °C (RTA), 1500 °C, and 1700 °C. While no significant emission peak from implanted Pr ions was found in as-implanted samples, two distinct peaks appeared in the visible wavelength range after thermal annealing. The most intense peak is at 527.1 nm, which is attributed to the ³P_J–³H₄ transition (J = 0, 1, 2). The second most intense peak is 655.7 nm (655.9 nm), which is thought to be attributed to the ³P₀–³F₂ [40–42] or ¹D₂–³H₄ transition [37,38,43], according to Dieke’s diagram. All emission peaks in PL were in good agreement with those in CL. A slight difference in peak wavelength between PL and CL is thought to be due to a misalignment of the grating in the spectrometer. Figure 2 (c) shows the PL spectrum of 100 keV Pr implanted GaN at the dose of 3 × 10¹⁴cm^-2 after thermal annealing at 1200 °C (RTA). In contrast to AlN, no emission peak due to the ³P_J–³H₄ transition appeared, although this transition appeared only at low temperatures [40] and can behave as a resonant excitation wavelength at room temperature [44]. The most intense emission peak in GaN is 652.4 nm, which has been attributed to the ³P₀–³F₂ transition [40–42], although it may be due to the ¹D₂–³H₄ transition, which is known to be the most dominant luminescence transition level in other materials such as YAG, LaF₃, and Y₄Al₂O₉ [37,38,45,46].

 

Fig. 2. (a) Room temperature PL spectra and (b) CL spectra of 100 keV Pr implanted AlN at the dose of 3 × 10¹⁴cm^-2 with different annealing temperatures: 1200 °C (red), 1500 °C (blue), and 1700 °C (pink). The excitation wavelength is 266 nm. The ordinate is normalized to the peak at 527.1 nm for the 1500 °C data. × 0.1 and ×10 offsets are applied to the 1200 °C and 1700 °C data for clarity. (c) PL spectrum of 100 keV Pr implanted GaN at the dose of 3 × 10¹⁴cm^-2 after annealing at 1200 °C (RTA). The ordinate is normalized to the value at 652.4 nm. (d) Intensity variation of the emission peak at 527.1 nm with annealing temperatures and implantation doses. The ordinate is normalized by the data for annealing at 1600 °C after implantation at 1 × 10¹⁵cm^-2.

Download Full Size | PDF

Luminescence centers in semiconductors are excited by free carrier recombination energy generated by electron beams and then emit photons in the relaxation process (i.e., non-resonant excitation or indirect excitation). Since 266 nm light (4.66 eV) exceeds the bandgap energy of GaN (3.4 eV), in addition to excitation to the 4f5d band, Pr ions can also be excited by the energy transfer of free carrier recombination generated in the GaN host [41,42]. Although the energy of 266 nm light is below the bandgap of AlN (6.0 eV), deep levels formed by point defects in the bulk and/or surface defects contribute to free carrier generation via two-photon excitation, and part of the implanted Pr ions could be non-resonantly excited by energy transfer from free carrier recombination. A similar phenomenon has been reported elsewhere [47]. It has been reported that oxygen impurities in AlN formed the optical absorption band in the energy range of 4–5 eV [48,49]. In addition, several point/complex defects generating photon absorption and emission have been reported: 2.8 eV associated with aluminum vacancies (V_Al) [50], 2.9–3.4 eV associated with complex defects of V_Al and silicon (Si) impurities (V_Al(Si_Al)_n) [51], 3.2–3.5 eV associated with complex defects of V_Al and oxygen (O) impurities (V_Al(O_N)_n) [50], and 3.9 eV associated with carbon impurities substituted at the N site (C_N) [49]. Free carriers can be generated in AlN by 266 nm light due to the contribution of these defects.

Although the similar PL spectra were obtained in both PL and CL after the annealing temperatures below 1500 °C, the PL and CL spectra changed significantly after annealing at 1700 °C, as shown in pink in Figs. 2(a) and (b). Two emission peaks at 527.1 nm and 655.9 nm were considerably reduced and new emission peaks at 482–484 nm and 605–635 nm appeared as shown by pink arrows in Fig. 2(a). This indicates the formation of different luminescence centers and/or quenching due to the Pr aggregation after annealing at 1700 °C. Since oxygen (O diffusion from the substrate occurs at 1700 °C [52], the drastic decrease in luminescence intensity and the appearance of another emission peak after the annealing at 1700 °C may be related to the O diffusion from the sapphire substrate.

Figure 2(d) shows the variation of the emission peak intensity at 527.1 nm as a function of annealing temperature at different implantation doses. The peak intensity increased with increasing annealing temperature for the implantation dose above 3 × 10¹⁴cm^-2, whereas no clear dependence was observed for the dose below 1 × 10¹⁴cm^-2. This trend is thought to be due to the different degree of implantation damage, which will be discussed in the next section.

Eu implanted AlN also shows strong luminescence under the 266 nm excitation, as shown in Fig. 3(a). The emission peak at 624.1 nm is attributed to the ⁵D₀–⁷F₂ transition, according to Dieke’s diagram [53–55]. The emission peaks due to the ⁵D₀–⁷F₂ transition appeared at 621.9 nm and 622.4 nm in the case of Eu implanted GaN, as shown in Fig. 3(b). The multiple peaks are caused by crystal filed splitting. It should be noted that the Eu implanted AlN showed a weak emission peak at 624.1 nm even before thermal annealing, unlike the case of Eu implanted GaN. This is probably caused by the dynamic annealing of AlN at room temperature [28,56]. The peak intensity at 624.1 nm decreased with increasing annealing temperature from 1400 °C to 1600 °C.

 

Fig. 3. (a) PL spectra of 100 keV Eu implanted AlN at the dose of 3 × 10¹³cm^-2 with different annealing temperatures: Pristine (dashed), as implanted (black), 1400 °C (gold), 1500 °C (blue), and 1600 °C (brown). The excitation wavelength is 266 nm. The ordinate is normalized to the peak value at 624.1 nm for the 1400 °C data. The PL spectrum of the pristine (unimplanted AlN) sample is also shown in black. (b) PL spectrum of 100 keV Eu implanted GaN at the dose of 3 × 10¹³cm^-2 after annealing at 1200 °C. The PL spectrum of the unannealed sample is also shown in black.

Download Full Size | PDF

3.2 Optical properties of implanted Pr and Nd ions under resonant excitation

Since the most intense emission peak of the Pr implanted AlN is 527.1 nm due to the ³P_J–³H₄ transition as shown in Fig. 2 (a), the implanted Pr ions can be directly excited by illuminating the 527.1 nm laser (resonant excitation). We first scanned the excitation laser wavelength by using a single frequency wavelength tunable laser from 522.2 nm to 527.2 nm and found that the maximum PL intensity was obtained when the excitation wavelength was 527.1 nm. Figure 4(a) shows the PL spectra of 100 keV-Pr implanted AlN at the dose of 1 × 10¹⁴cm^-2 under resonant excitation. The peak at 655.2 nm due the ³P₀–³F₂ or ¹D₂–³H₄ transition was also observed in the as implanted (unannealed) sample. A peak at 694 nm is thought to be due to the Chromium (Cr) emission in sapphire substrate since it was unaffected by Pr implantation and annealing. After thermal annealing, not only the peak at 655.2 nm, but also multiple peaks appeared at 670–700 nm, 750–770 nm, 780–820 nm regions due to the 4f-4f transitions of Pr³⁺ ions (i.e., ³P–³F, ³P–³H, and ¹D–³H transitions). No spectral change was observed after annealing up to 1500 °C, but most of the emission peaks disappeared after annealing at 1700 °C, as is the case in Fig. 2(a), although no new emission peak was not found in the measured wavelength range. Figure 4(b) shows the intensity variation of the emission peak at 655.9 nm due to annealing temperatures with different implantation doses. The emission intensity increased with increasing temperature in the case of 1 × 10¹⁵cm^-2 but decreased when the dose was below 1 × 10¹⁴cm^-2. This opposite trend will be discussed in the next section.

Figure 5 shows the PL spectra of Nd implanted AlN and GaN under 785 nm laser excitation. Since a broad resonant excitation band at 760–800 nm exists in Nd implanted GaN due to the ⁴I_9/2–⁴F_7/2, ⁴S_3/2 transition [57], it would appear that the similar resonant excitation also exists at 785 nm in Nd implanted AlN. The most intense peak appeared at 914.7 nm in AlN and 916.3 nm in GaN is attributed to the ⁴F_3/2–⁴I_9/2 transition. The as implanted (unannealed) AlN also showed the emission peak at 914.7 nm, suggesting that the significant dynamic annealing at room temperature contributed to the optical activation of implanted Nd ions. The Nd-related emission peaks decreased with increasing temperature from 1400 °C to 1600 °C, but new emission peaks at 880 to 900 nm appeared after the annealing at 1600 °C. These peaks also appeared slightly at 1400 °C and 1500 °C, indicating that the implanted Nd ions converted to other luminescence centers at high temperature.

 

Fig. 4. (a) PL spectra under resonant excitation of 100 keV Pr implanted AlN at the dose of 1 × 10¹⁴cm^-2 with different annealing temperatures: As implanted (black), 1200 °C (red), 1500 °C (blue), and 1700 °C (pink). The excitation wavelength is 527.1 nm. The ordinate was normalized to the peak at 655.2 nm for the 1200 °C data. The PL spectrum of the unannealed sample is also shown in black. (b) Intensity variation of the emission peak at 655.2 nm with annealing temperature and implantation dose. The ordinate is normalized to the value of the annealing at 1600 °C annealing for 1 × 10¹⁵cm^-2 implantation.

Download Full Size | PDF

 

Fig. 5. (a) PL spectra of 100 keV Nd implanted AlN at the dose of 3 × 10¹³cm^-2 with different annealing temperatures: Pristine (dashed), as implanted (black), 1400 °C (gold), 1500 °C (blue), and 1600 °C (brown). The excitation wavelength is 785 nm. The ordinate is normalized to the peak at 914.7 nm for the 1400 °C data. The PL spectrum of the pristine (unimplanted AlN) sample is also shown in black. (b) PL spectrum of 100 keV Nd implanted GaN at the dose of 3 × 10¹³cm^-2 after annealing at 1200 °C.

Download Full Size | PDF

The luminescence lifetime of implanted Pr and Nd ions was characterized using the TCSPC system with pulsed lasers. Figure 6 shows the time-resolved PL spectra and the obtained lifetimes. In the TRPL measurement for Pr ions, almost all of the photon emission from the sample was attributed to the implanted Pr ions because the excitation wavelength (527.1 nm) matched to the energy levels between the ³P_J and ³H₄ state, and the 655 nm bandpass filter with 40 nm band was placed in front of the APD to remove signals from other luminescent components such as bulk/surface defects and other impurities. As a result, the other luminescent components were negligibly small, and thus the TRPL spectrum was fitted by the single exponential decay function ($A\exp ({ - t/{\tau_{\textrm{Ln}}}} )+ B$, where t is the time and $\tau_{\textrm{Ln}}$ is the lifetime of the Ln ions). The TRPL spectra of as-implanted and 1500 °C annealed Pr-implanted AlN samples are shown in Fig. 6(a). The fitted curves are also plotted in blue. The obtained luminescence lifetimes as a function of annealing temperature are shown in Fig. 6(b). In the case of 100 keV-Pr ions at 1 × 10¹⁴cm^-2, the lifetime was reduced from 11 µs to 8.1∼8.7 µs by thermal annealing. No clear trend was found in the temperatures between 1200 °C and 1500 °C. The similar trend was found from the results of 700 keV-Pr ions at 3 × 10¹³cm^-2 and 100 keV-Pr ions at 1 × 10¹⁵cm^-2. The longer lifetime was obtained for 700 keV-Pr ions (10.3∼10.5 µs) than for 100 keV-Pr ions (8.1∼8.7µs). This result suggests that a non-radiative transition due to surface defects significantly affected the relaxation process of excited Pr ions, because the ion ranges of 100 keV- and 700 keV-Pr ions were 36 nm and 186 nm from the surface, respectively (see Fig. 1). Moreover, these values are significantly shorter than those of Pr ions implanted in GaN, whose lifetime has been reported to be about 14 µs [58], suggesting that the contribution of the non-radiative transition is larger in AlN than in GaN.

 

Fig. 6. (a) TRPL spectra of 100 keV-Pr implanted AlN at 1 × 10¹⁵cm^-2 (as implanted and annealed at 1500 °C). (b) Lifetime variations of implanted Pr in AlN with annealing temperature and implantation condition. (c) TRPL spectra of 700 keV-Nd implanted AlN at 3 × 10¹³cm^-2 (as implanted and annealed at 1500 °C). (d) Lifetime variations with annealing temperature. Fast decay components are attributed to defect related luminescence (green and blue triangles).

Download Full Size | PDF

Figure 6(c) shows the TRPL spectra of Nd ions implanted in AlN. The excitation wavelength was 785 nm and a 900 nm longpass filter was placed in front of the APD to reduce unwanted luminescence in the sample. In this measurement condition, a considerable number of photons from luminescence centers other than Nd ions were still included in the spectrum because the excitation wavelength did not completely match the excitation transition from the⁴I_9/2 state to the ⁴F_7/2 or ⁴S_3/2 state and the other components were relatively high. The obtained TRPL spectra clearly contained at least two components: the fast decay component and the slow decay component. Therefore, the TRPL spectrum after annealing was fitted by the double exponential decay function (${A_1}\exp ({ - t/{\tau_{\textrm{other}}}} )+ {A_2}\exp ({ - t/{\tau_{\textrm{Ln}}}} )+ B$, where ${\tau _{\textrm{other}}}$ is the lifetime of components other than Nd ions). For the TRPL spectrum of the as implanted (unannealed) sample, the triple exponential decay function was used for a better fit (${A_1}\exp ({ - t/{\tau_{\textrm{other1}}}} )+ {A_2}\exp ({ - t/{\tau_{\textrm{other2}}}} )+ {A_3}\exp ({ - t/{\tau_{\textrm{Ln}}}} )+ B$), since the double exponential decay function did not fit the experimental data well, as shown by a green curve in Fig. 6(c). The luminescence lifetimes obtained from these fits are shown in Fig. 6(d). The fast component (2.1∼2.6 µs) was unchanged by thermal annealing, while the slow component was extended from 83.4 µs to 133.1∼141.1 µs. Since the reported lifetime of implanted Nd ions in GaN is about 110 µs [57], it can be concluded that the slow component is attributed to implanted Nd ions in AlN. It is known that the lifetime is extended by thermal annealing because the nonradiative recombination due to energy transfer to neighboring defects is reduced ($1/{\tau _{\textrm{exp}}} = 1/{\tau _{\textrm{PL}}} + 1/{\tau _{\textrm{NR}}}$, where ${\tau _{\textrm{exp}}}$ is the experimentally obtained lifetime, ${\tau _{\textrm{PL}}}$ is the actual luminescence lifetime, and ${\tau _{\textrm{NR}}}$ is the nonradiative transition lifetime). The change in lifetime of implanted Nd ions is expected to reflect such effects, although the change in lifetime of implanted Pr ions showed the opposite trend.

3.3 Implantation damage and recovery after thermal annealing

To analyze the ion implantation damage in AlN crystal and its recovery by thermal annealing, XRD and AFM studies were performed. In addition, the thermal diffusion of the implanted Pr ions in AlN after thermal annealing was investigated by SIMS. Figure 7(a) shows the XRD 2θ-ω profiles around AlN (0002) of 100 keV-Pr ion implanted AlN at the dose of 3 × 10¹⁴cm^-2 and the evolution of the spectrum with annealing temperatures. The spectrum of the pristine sample (unimplanted AlN) is also shown as a black line and the peak at 35.97 degrees is attributed to AlN (0002). After implantation, the peak intensity decreased remarkably and new satellite peaks appeared at 35.66 and 35.80 degrees, although the FWHM of the 2θ-ω profiles was unchanged. The satellite peaks are attributed to lattice expansion in the Pr implanted region, while the unchanged center peak at 35.97 degrees is attributed to the signal from the unimplanted region, which is deeper than the Pr implanted region (∼50 nm below the surface). Similar results have been reported elsewhere [59,60]. The shoulder peak at the lower angle side, which was clearly visible after annealing at 1200 °C, decreased with increasing annealing temperature. This indicates that the lattice expansion due to the accumulation of implantation induced defects was reduced with increasing annealing temperature. Figure 7(b) shows the change in the intensity ratio of the AlN (0002) peak (35.94–36.00 degrees) to the total intensity (35.50–36.30 degrees). The intensity ratio degraded due to implantation increased with increasing annealing temperature, although the value after annealing at 1700 °C was still slightly lower than the value before implantation (pristine) due to the residual implantation induced damage.

 

Fig. 7. (a) XRD 2θ-ω spectra of 100 keV-Pr ions implanted AlN at different annealing temperatures: Pristine (black), as implanted (black), 1200 °C (red), 1300 °C (green), 1500 °C (blue), 1700 °C (pink). The spectra of unimplanted (pristine) and as implanted samples are also shown. The ordinate is normalized to the peak value at 35.97 degrees, and a × 100 offset is applied to each spectrum for clarity. (b) The intensity ratio of the peak at 35.97 degrees (35.94–36.00 degrees marked as yellow in (a)) to the total intensity (35.50–36.30 degrees).

Download Full Size | PDF

The AFM images are shown in Fig. 8. As shown in Fig. 8(a), a fine atomic step-terrace structure was clearly visible at the surface before implantation [35] and the root mean square (RMS) roughness was estimated to be 0.08 nm. The RMS was degraded to be above 0.15–0.18 nm after 100 keV-Pr ion implantation. The atomic step-terrace structure disappeared, and the island structure appeared. This is probably due to the sputtering effect of Pr ions. However, no significant change in RMS was found with increasing Pr implantation dose under these irradiation conditions (Figs. 8(b)-(e)). This result suggests that the surface roughness degradation was saturated at the low dose (1 × 10¹²cm^-2) and the further ion implantation did not cause any change in the surface roughness, since the ion track overlap starts to occur above the dose of about 10¹²cm^-2 [61]. After thermal annealing, the island structure disappeared, and the atomic step-terrace structure reappeared (Figs. 8(f)-(h)). This is believed to be due to the removal or recrystallization of the surface amorphous layer. The linearity of the atomic step-terrace structure improved with increasing annealing temperature, indicating that the surface recovery was obtained at higher temperatures.

Figure 9 shows the SIMS profiles of implanted Pr ions in AlN at different annealing temperatures. The implantation energy and dose were 100 keV and 3 × 10¹⁴cm^-2, respectively. Figure 9(b) shows the magnified profiles of the same data as in Fig. 9(a). The 100 keV-Pr ion profile calculated by TRIM is also shown in Fig. 9(b). Although the implantation angle was 7 degrees inclined from the c-axis of AlN, some of the Pr ions were implanted deeper than the calculated ion range due to the channeling effect. The peak depth was 31 nm from the surface, which was slightly shallower than the calculated value (36 nm) by TRIM. Although no significant change of the Pr ion profile was found after annealing at 1200 °C (RTA), thermal diffusion of Pr ions appeared at the temperature above 1300 °C. As shown in Fig. 9(a), the diffused Pr ions accumulated at the interface of MOVPE AlN and sputter AlN with increasing temperature, and the Pr diffusion reached to the interface of sapphire substrate after annealing at 1700 °C. The Pr ions also diffused to the surface at the temperature above 1300 °C, as clearly found in Fig. 9(b). This result suggests that the deactivation (quenching) of the implanted Pr ions due to the aggregation at the interface was enhanced with the increase of annealing temperature. Interestingly, almost no Pr remained in the sputter-AlN layer. Although the cause of this is not clear at this time, a high concentration of O impurity in the sputter-AlN layer (above 10²⁰cm^-3) [35] may have negatively affected the Pr stabilization in AlN.

 

Fig. 8. AFM images of 100 keV-Pr implanted AlN at different doses and annealing temperatures. (a) pristine (unimplanted AlN), (b) implanted AlN at the dose of 1 × 10¹²cm^-2 without annealing, (c) 1 × 10¹³cm^-2 without annealing, (d) 1 × 10¹⁴cm^-2 without annealing, (e) 3 × 10¹⁴cm^-2 without annealing, (f) 3 × 10¹⁴cm^-2 with annealing at 1300 °C, (g) 3 × 10¹⁴cm^-2 with annealing at 1500 °C, (h) 3 × 10¹⁴cm^-2 with annealing at 1700 °C. The measured area was 1 × 1 µm² and the obtained RMS is shown at the bottom of the figure.

Download Full Size | PDF

 

Fig. 9. (a) SIMS profiles of 100 keV-Pr implanted AlN at 3 × 10¹⁴cm^-2 at different annealing temperatures: As implanted (black), 1200 °C (red), 1300 °C (green), 1500 °C (blue), and 1700 °C (pink). (b) The enlarged spectra up to 250 nm from the surface. The implanted Pr ion profile simulated by TRIM is shown in gray. A ×10 offset is applied to each data for clarity.

Download Full Size | PDF

4. Discussion

In this section, we discuss the relationship between the optical activation of implanted Pr ions in AlN and the recovery of implantation damage in the crystal after thermal annealing at different temperatures. The results of XRD and AFM studies shown in the subsection 3.3 clarified that the implantation induced damage was recovered at higher temperatures up to 1700 °C. On the other hand, the PL intensity of implanted Pr ions did not show a monotonic change with annealing temperature (Figs. 2 and 4). The PL intensity increased with increasing temperature when 100 keV-Pr ions were implanted at the dose above 3 × 10¹⁴cm^-2, whereas the PL intensity decreased with increasing temperature under the resonant excitation and no clear trend was found under the 4f5d band excitation when the implantation dose was below 1 × 10¹⁴cm^-2. However, after annealing at 1700 °C, the PL intensity decreased drastically and new emission peaks appeared in all the cases. The similar change was found in the PL of Nd implanted AlN as shown in Fig. 5.

Implanted Pr ions are stabilized as luminescence centers by substitution into cation sites in the trivalent state (Pr³⁺) [11]. Post thermal annealing is necessary to restore the radiation induced damage and stabilize Pr ions as luminescence centers. Neighboring defects and damage regions can cause nonradiative relaxation processes of the excited Pr ions, and thus recovery of crystallinity by thermal annealing is generally an important factor for highly efficient activation of implanted Pr ions. However, the post thermal annealing can also cause the complex defects of Pr ions with other impurity defects and vacancies, resulting in the deactivation (quenching) or change of the emission spectrum. In addition, thermal diffusion of Pr ions to the surface and interfaces can induce aggregation and/or formation of another compound. Therefore, the PL properties of implanted Pr ions after thermal annealing are determined by the trade-off between the optical activation of Pr³⁺ associated with the recovery of radiation induced damage and the deactivation due to complex defect formation and aggregation of Pr ions.

The damage buildup due to ion implantation and the recovery by thermal annealing can be discussed in terms of the displacement damage dose (dpa). The lattice disorder increases due to the accumulation of implantation induced defects, and the crystal lattice is eventually amorphized. According to the K. Lorenz, et al. [28] and W. Jiang, et al. [62], the lattice disorder of AlN starts to occur when the implantation dose exceeds about 1 dpa, and the severe implantation damage (i.e., amorphization) appears at 10∼100 dpa. Here, we define the critical dose as the dose at which the lattice disorder starts to occur. It is considered difficult to recover from implantation induced damage when the implantation dose exceeds the critical dose. This value roughly corresponds to the implantation dose of 3 × 10¹⁴cm^-2 for 100 keV-Pr ions, according to the TRIM calculation (see Fig. 1(b)). The displacement energy used for the calculation was 25 eV for Al and 28 eV for N, which are the default values in TRIM, although different values have been predicted [63,64]. This fact suggests that above the critical dose, a higher annealing temperature is required to obtain a better recovery of implantation damage due to the partial amorphization of the AlN lattice, and to obtain a better optical activation of the implanted Pr ions. In this case, the increase in quenching factors at high temperatures is thought to be less significant, and thus the PL intensity increased with increasing annealing temperature by 1600 °C (Figs. 2(d) and 4(b)). On the other hand, since the effects of implantation damage recovery on the Pr activation are less significant when the implantation dose is below the critical dose, the PL intensity under the resonant conditions decreased with increasing annealing temperature due to the increased in quenching factors.

5. Conclusion

We investigated the optical properties of Pr, Eu, and Nd ion implanted AlN semiconductors and their annealing temperature dependence. The dominant luminescence transition for implanted Pr ions in AlN was different from that in GaN. The emission peak at 527.1 nm due to the ³P_J–³H₄ transition of Pr³⁺ ions was dominant in AlN, while the emission peak at 652.4 nm due to the ³P₀–³F₂ or ¹D₂–³H₄ transition was dominant in GaN. The dominant luminescence transitions for implanted Eu and Nd ions did not change depending on the host material. The emission peak appeared at 914.7 nm for AlN and 916.3 nm for GaN was attributed to the ⁴F_3/2–⁴I_9/2 transition of Nd³⁺ ions. Also, the emission peak appeared at 624.1 nm for AlN and 621.9 nm and 622.4 nm for GaN was attributed to the ⁴F_3/2–⁴I_9/2 transition of Eu³⁺ ions.

The XRD and AFM studies showed that the implantation damage recovered with increasing annealing temperature and was not completely eliminated even by annealing at 1700 °C. Due to the recovery of AlN crystallinity, the PL intensity of implanted Pr ions increased with increasing annealing temperature by 1600 °C when the implantation dose was high (above 3 × 10¹⁴cm^-2 in the case of 100 keV), in other words, the implantation damage was high (above 1 dpa). When the implantation dose was low (below 1 dpa), however, no significant enhancement of Pr³⁺ photon emission under the excitation to the 4f5d band was found due to the increase in annealing temperature. In addition, the PL intensity decreased under resonant excitation condition, which is the opposite trend for the case of high dose implantation. This complicated annealing behavior can be explained by the increase in quenching factors such as complex defect formation and aggregation at the surface and interfaces at high temperatures. The SIMS analysis clarified that the thermal diffusion of Pr ions and the aggregation at the interface occurred at the temperature above 1300 °C, and the Pr ions reached the interface on the sapphire substrate at 1700 °C. The PL intensity decreased significantly after annealing at 1700 °C regardless of the implantation dose, and other new peaks appeared, indicating the formation of Pr complex defects. Therefore, it is concluded that the optical activation of implanted Pr ions in AlN by thermal annealing is determined by the balance between the recovery of implantation damage and the increase of quenching factors such as complex defect formation and aggregation. This trade-off relationship was demonstrated experimentally because AlN is the host material that can withstand high temperatures up to at least 1700 °C. The similar trade-off relationship regarding the optical activation of implanted luminescence centers may be found in the other III-N materials such as GaN.