Introduction

Photoluminescence (PL) and photoluminescence excitation (PLE) spectroscopy have been performed to study the different structural configurations (sites or centers) on which Er³⁺ ions are incorporated in Er-doped GaN and to investigate their excitation mechanisms [Reference Kim, Rhee, Turnbull, Reuter, Li, Coleman and Bishop1-Reference Hansen, Zhang, Perkins, Safvi, Zhang, Bray and Kuech7]. Our previous study of site-selective PL and PLE spectroscopy in Er-implanted samples of GaN grown by metal organic chemical vapor deposition (MOCVD) that were annealed at 900°C under a flow of N₂ revealed the existence of seven different Er³⁺ sites [Reference Kim, Rhee, Li, Coleman and Bishop3]. Six of these Er³⁺ PL sites are attributed to complexes of Er atoms with defects and impurities. Only one of the seven sites can be pumped by direct 4f absorption; the concentrations of the other sites are too low to allow excitation by direct 4f absorption. These results raise an obvious question concerning the generality or uniqueness of the Er³⁺ centers observed in GaN. That is, are these multiple, selectively excited, discrete Er³⁺ sites or centers unique to Er-implanted MOCVD-grown GaN subjected to specific post-implantation annealing conditions, or are they also present in Er-doped GaN synthesized by other growth, doping and annealing procedures?

Site-selective PL and PLE spectroscopies used in our previous work are applied in this study to Er-implanted hydride vapor phase epitaxy (HVPE)-grown GaN annealed at 800°C in an NH₃/H₂ atmosphere and in situ Er-doped HVPE-grown GaN [Reference Hansen, Zhang, Perkins, Safvi, Zhang, Bray and Kuech7] to see if any of the seven different Er³⁺ sites observed in Er-implanted MOCVD-grown GaN exist in these samples. The PLE and PL of these two different HVPE-grown samples are compared here first with the Er-implanted MOCVD-grown sample and then the differences and similarities among the site-selective PLE and PL spectra of these three different Er-doped GaN films are discussed in detail.

Experimental Procedure

The GaN films were doped in-situ with Er in a horizontal HVPE reactor during growth. A peak Er concentration of 2 × 10¹⁹ ions/cm³ was achieved in this in situ Er-doped GaN at a thickness of 1000 nm [Reference Hansen, Zhang, Perkins, Safvi, Zhang, Bray and Kuech7]. The GaN films grown on sapphire by HVPE were implanted with a dosage of 2 × 10¹⁴ ions/cm² at 300 keV. The peak concentration of Er is 5.3 × 10¹⁹ ions/cm³ at a depth of 33 nm [Reference Hansen, Zhang, Perkins, Safvi, Zhang, Bray and Kuech7]. These Er-implanted HVPE-grown GaN films were annealed in a conventional tube furnace at 800°C for 30 minutes in a flowing NH₃/H₂. For comparison, the GaN films grown on sapphire by atmospheric pressure MOCVD were implanted with a dosage of 4 × 10¹³ ions/cm² at 280 keV [Reference Kim, Rhee, Turnbull, Reuter, Li, Coleman and Bishop1-Reference Kim, Rhee, Li, Coleman and Bishop4]. The peak concentration of Er is 2 × 10¹⁸ ions/cm³. Post-implantation annealing was carried out in a conventional tube furnace at 900°C for 30 minutes under a continuous flow of nitrogen gas.

6K PL spectroscopy was carried out on the three different Er-doped GaN samples. The PL spectra were excited by a variety of sources including a tunable titanium-doped sapphire laser, a HeNe laser, an Ar ion laser, a Xe lamp dispersed by a double grating monochromator, and a HeCd laser. The PLE spectra were obtained with a xenon lamp dispersed by a double grating monochromator or with a tunable titanium-doped sapphire laser. All of the PLE spectra were corrected for the spectral response of the tunable excitation systems. The luminescence was analyzed by a 1-m single grating monochromator and detected by a cooled Ge PIN detector. Samples were cooled to liquid helium temperature in a Janis Supervaritemp Cryostat.

Results and Discussion

Figure 1 shows the PL spectra obtained at 6 K under excitation by 515 nm light (“green-pumped”) from in situ Er-doped and Er-implanted GaN grown by HVPE. The PL spectrum taken for Er-implanted MOCVD-grown GaN under the same experimental conditions is also shown for comparison. All three PL spectra exhibit the 1540 nm band characteristic of the ⁴I_13/2 → ⁴I_15/2 transitions of Er³⁺ and broad background PL bands on which the Er-related PL bands are superimposed. In the PL of the in situ Er-doped GaN (Fig. 1a), the broad PL bands have different lineshapes and peak positions from those of the damage-induced broad-band PL observed in the Er-implanted MOCVD-grown GaN (Fig. 1c) [Reference Kim, Rhee, Li, Coleman and Bishop4]. Since these bands have not been observed in undoped HVPE-grown GaN, they are apparently induced by the in situ doping during growth. In the PL spectrum of the Er-implanted HVPE-grown GaN (Fig. 1b), the damage-induced broadband PL is barely observable.

Fig. 1. The Er³⁺ PL spectra and the broad defect PL bands (pumped by 515 nm light).

Figure 2 displays the PLE spectra of the three different Er-doped samples obtained by detecting the integrated Er³⁺ PL intensity while scanning the wavelength of the xenon lamp-double monochromator excitation system. The PLE spectrum 2(a) of the in situ Er-doped sample exhibits relatively narrow absorption bands attributable to direct optical excitation of the intra 4f-shell transitions of Er³⁺, and an exponential absorption tail just below the 3.5 eV band gap. The exponential energy dependence of the below-gap absorption tail is reminiscent of the Urbach absorption edge that characterizes the band edge of disordered materials such as chalcogenide glasses [Reference Turnbull and Bishop8] or highly doped crystalline semiconductors [Reference Mott and Davis9]. This PLE spectrum does not show the broad, mid-gap defect- or impurity-related absorption bands that are observed in the PLE spectra 2(b) and 2(c) of the Er-implanted samples. The absorption peaks seen at 378, 400, 445, 487, 521, 543, 651, 796, and 970 nm in the PLE spectrum 2(a) are attributed to ⁴I_15/2 → ⁴G_11/2 ⁴I_15/2 → ⁴H_9/2 ⁴I_15/2 → ⁴F_3/2,5/2 ⁴I_15/2 → ⁴F_7/2, ⁴I_15/2 → ⁴H_11/2, ⁴I_15/2 → ⁴S_5/2, ⁴I_15/2 → ⁴F_9/2, ⁴I_15/2 → ⁴I_9/2,and ⁴I_15/2 → ⁴I_11/2 Er³⁺ intra 4f-shell transitions, respectively [Reference Thaik, Hommerich, Schwartz, Wilson and Zavada5,Reference Wu, Hommerich, MacKenzie, Abernathy, Pearton, Schwartz, Wilson and Zavada6,Reference Gan10]; some of these PLE bands were reported in Refs. 5 and 6 to be observed in Er-implanted GaN and in situ Er-doped AlN grown by molecular-beam epitaxy (MBE). In contrast, the PLE spectra 2(b) and 2(c) obtained from the Er-implanted HVPE- and MOCVD-grown samples, respectively, show only the broad, defect- or impurity-related absorption bands that are not attributable to Er³⁺. These absorption bands are characteristic of the implanted films only, suggesting that some of the broad absorption bands in the PLE spectra 2(b) and 2(c) are associated with defects or defect-impurity complexes created during the implantation and annealing procedures [Reference Kim, Rhee, Turnbull, Reuter, Li, Coleman and Bishop1,Reference Kim, Rhee, Li, Coleman and Bishop3].

Fig. 2. The PLE spectra obtained by detecting the integrated Er³⁺ PL intensity.

The high-resolution PLE spectra (Fig. 3) obtained in the 775−825 nm spectra range with the tunable Ti-doped sapphire laser exhibit PLE peaks assigned to direct ⁴I_15/2 → ⁴I_9/2 4f shell absorption. The absorption bands seen in the PLE spectrum (Fig. 3a) of the in situ Er-doped sample are much broader and their peak positions are shifted to the shorter wavelength region, compared to the Er-implanted samples. Note that the PLE spectrum of the MOCVD-grown GaN in Fig. 3c shows five pairs of sharp peaks, indicative of a single type of Er³⁺ center or site. The loss of the sharp structure in the PLE spectrum of the in-situ Er-doped GaN sample indicates that the Er dopants in this sample occupy sites located in disordered regions of the material. The disordered or “amorphous” character of the Er³⁺ centers’ structural environments gives rise to site-to-site variations in the magnitude and symmetry of the crystal fields experienced by the Er³⁺ ions that are responsible for the spectral broadening of the Er³⁺ 4f PLE bands [Reference Turnbull and Bishop8,Reference Mott and Davis9]. Additional evidence for disorder in the structural environment of the Er dopants in the GaN doped during growth is provided by the broad Urbach absorption edge in the near-band edge PLE spectrum of Fig. 2a, which contrasts strongly with the relatively sharp band edge absorption features observed in the PLE spectra of the Er³⁺ emission in Er-implanted samples of GaN (see, for example, Fig. 2c). Furthermore, the 1540 nm Er³⁺ PL bands from the in situ Er-doped sample shown in Figs. 1a, 5a, andFig7 7a all have a broadened line shape more characteristic of Er dopants in glasses [Reference Turnbull and Bishop8], than the sharply structured 1540 nm Er³⁺ PL bands of the Er-implanted GaN samples shown inFigs. 1b and c, 5b and c, 6a and b, and 7b and c.

Fig. 3. The ⁴I_15/2 → ⁴I_9/2 4f shell Er³⁺ PLE _spectra.

The PLE spectrum of the Er-implanted HVPE-grown GaN in Fig. 3b exhibits sharp-structured peaks having the same peak positions as those in the Er-implanted MOCVD-grown GaN (Fig. 3c). A careful comparison between these two PLE absorption bands reveals that the PLE peaks in the HVPE-grown one are broader, compared to the MOCVD-grown one. The broadening of the sharp ⁴I_15/2 → ⁴I_9/2 4f shell absorption peaks in the Er-implanted, HVPE-grown sample is apparently attributable to the fact that the concentration (5.3 × 10¹⁹ ions/cm³) of the implanted Er³⁺ ions is larger than that (2 × 10¹⁸ ions/cm³) in the Er-implanted MOCVD-grown sample; a similar broadening of the sharp absorption peaks was observed in the PLE spectrum of MOCVD-grown GaN implanted with 10¹⁵ Er ions/cm² at 350 keV reported in Ref. 11. While the sharpness of the PLE peaks in the PLE spectrum of the Er-implanted MOCVD-grown GaN indicates that isolated Er ions on Ga atomic positions are on identical, high-quality sites in this material [Reference Kim, Rhee, Li, Coleman and Bishop4], the spectral broadening of the direct 4f PLE absorption peaks for the Er-implanted HVPE-grown sample implies moderate site-to-site variation in the crystal field.

The seven distinct optically active Er³⁺ centers observed previously in the Er-implanted MOCVD-grown GaN annealed at 900°C (Ref. 3) have been labeled on the basis of their excitation wavelengths, as follows: above-gap (A), blue (B), orange (O), red (R), near-IR (N), 4f (F) and violet (V). This labeling system has been used to identify the PLE absorption bands associated with the selective excitation of these Er³⁺ centers in the PLE spectra in Figs. 2c and 3c. Figures 4, 5, and 6 show the “violet-pumped”, “blue-pumped”, and “red-pumped” PL spectra excited by 404, 458 and 633 nm light, respectively, that correspond with the similarly labeled PLE bands in Fig. 2c [Reference Kim, Rhee, Li, Coleman and Bishop4]. The “violet-pumped” PL spectra presented in Fig. 4 demonstrate that the PL peaks in the “violet-pumped” PL spectrum of the Er-implanted HVPE-grown GaN are the same as those in the Er-implanted MOCVD-grown GaN, while there was no Er³⁺ emission from the in situ Er-doped GaN when excited by 404 nm (violet) light. In Fig. 5, the broadened PL peaks in the “blue-pumped” PL spectrum of the in-situ Er-doped sample do not correspond to the B-labeled PL peaks associated with the “blue” Er³⁺ site observed in the Er-implanted MOCVD-grown sample, but the PL of the Er-implanted HVPE-grown sample includes both the B- and A-labeled PL peaks. The “red-pumped” PL spectra in Fig. 6 show that while the 633 nm light excites the R (red)-labeled Er³⁺ emission as well as the B-, A-, N-, and O-labeled bands in the Er-implanted MOCVD GaN sample (Fig. 6b), only the B-labeled PL peaks are observable in the PL of the Er-implanted HVPE-grown sample (Fig. 6a), and no ∼1540 nm emission is excited from the in situ Er-doped GaN.

Fig. 4. The “violet-pumped” PL spectra as excited by violet light (404 nm).

Fig. 5. The “blue-pumped” PL spectra as excited by blue light (458 nm).

Fig. 6. The “red-pumped” PL spectra as excited by red light (633 nm).

The “4f-pumped” PL spectra in Fig. 7 were excited by 809 nm light that corresponds to direct Er ³⁺ 4f-band absorption. Significantly, the “4f-pumped” PL spectrum of the Er-implanted HVPE-grown GaN (Fig. 7b) shows sharply-structured PL peaks that are nearly identical in peak energy and relative intensities to those of the Er-implanted MOCVD-grown GaN PL spectrum of Fig. 7c. This indicates that most of the Er atoms in the Er-implanted HVPE-grown GaN occupy a single type of high-concentration, isolated Er³⁺ center, that can only be excited by direct 4f-band absorption, and that this is the same high-concentration Er³⁺ center previously observed by direct 4f-band pumping in Er-implanted MOCVD-grown GaN [Reference Kim, Rhee, Li, Coleman and Bishop3,Reference Kim, Rhee, Li, Coleman and Bishop4]. In contrast, 4f pumping of the Er³⁺ dopants in the in-situ Er-doped GaN excites a broad 1540 nm PL spectrum (Fig. 7a) that is identical to the “green-pumped” PL spectrum (Fig. 1a) and the “blue-pumped” PL spectrum (Fig. 5a) for this sample, clearly demonstrating the absence in the in-situ-doped material of the multiple, selectively excited, discrete Er³⁺ sites or centers that are observed in the Er-implanted GaN. The broad 4f-pumped Er³⁺ PL spectrum confirms our earlier suggestion that the Er dopants in the in-situ-doped sample occupy sites located in disordered regions of the material and therefore experience site-to-site variations in the crystal field.

Fig. 7. The “4f-pumped” PL spectra as excited by 809 nm light.

Conclusions

Site-selective PLE and PL spectroscopy have been carried out at 6K on the 1540 nm ⁴I_13/2 → ₄I_15/2 emission of Er³⁺ in in situ Er-doped and Er-implanted GaN grown by HVPE. The PLE and PL spectra of these two different Er-doped HVPE-grown GaN films are compared in this study with Er-implanted GaN grown by MOCVD to see if the multiple, selectively excited, discrete Er³⁺ sites or centers observed in our previous studies of Er-implanted GaN are present in Er-doped GaN synthesized by other growth, doping and annealing procedures.

The PLE and PL spectroscopy of this study reveal that four of the seven different Er³⁺ sites observed in Er-implanted MOCVD-grown GaN annealed at 900°C under a flow of N₂ are present in Er-implanted HVPE-grown GaN annealed at 800°C in an NH₃/H₂ atmosphere. In contrast, the in situ Er-doped HVPE-grown GaN exhibits a single, broad ∼1540 nm Er³⁺ PL spectrum whose lineshape is independent of excitation wavelength. This broad Er³⁺ PL spectrum, which is reminiscent of emission from an Er-doped glass, prompts the suggestion that the Er dopants in the in-situ-doped sample occupy sites located in disordered regions of the material characterized by site-to-site variations in the crystal field.