Introduction

The luminescence from rare earth doped III-nitrides is of significant current interest for potential applications in optical communications and full color displays.[Reference Steckl and Zavada1] Visible and infrared electroluminescence (EL) has been reported from a number of rare earth doped GaN systems: GaN: Er (green, IR)[Reference Torvik, Feuerstein, Pankove, Qui and Namavar2,Reference Steckl and Birkhahn3,Reference Steckl, Garter, Birkhahn and Scofield4], GaN: Pr (red, IR) [Reference Chao and Steckl5], GaN: Eu (red) [Reference Heikenfeld, Garter, Lee, Birkenhahn and Steckl6,Reference Hara and Ohtake7] and GaN: Tm (blue) [Reference Steckl and Zavada1] The incorporation, optical activation, and luminescence efficiency of rare earth ions in III-nitrides, however, is not yet fully understood. In this paper, we present a comparison of the PL properties of Er doped GaN grown by MOMBE and SSMBE. Excitation wavelength and temperature dependent PL studies were performed and analyzed in view of optoelectronic applications of Er doped GaN.

Experimental Procedures

The GaN: Er (MOMBE) sample was grown in an INTEVAC Gas Source Gen II on In-mounted (100) Si substrate as described in reference [Reference MacKenzie, Abernathy, Pearton, Hömmerich, Seo, Wilson and Zavada8]. The GaN: Er (SSMBE) sample was grown on Si by solid source and RF-assisted molecular beam epitaxy (MBE). Details of the Riber MBE32 system used for growth have been discussed previously [Reference Steckl and Birkhahn3]. PL studies were performed using a HeCd laser operating at either 325 nm or 442 nm. Infrared PL spectra were recorded using a 1-m monochromator equipped with a liquid-nitrogen cooled Ge detector. In visible PL studies a thermo-electric cooled PMT was employed for detection. The signal was processed using lock-in techniques. IR lifetime studies employed the 355 nm line of a Nd: YAG laser for excitation. Visible lifetime data were taken by pumping into the ⁴F_7/2 Er³⁺ transition at ∼495 nm.

Results and Discussion

The infrared luminescence spectra of Er doped GaN grown by MOMBE and SSMBE are shown in Figure 1 for above (λ_ex=325 nm) and below-gap excitation (λ_ex=442 nm). The spectra were taken under identical experimental conditions. The pump power was kept constant at ∼0.64 W/cm².

Figure 1: 1.54 µm PL spectra of GaN: Er (MOMBE) and GaN: Er (SSMBE) at room temperature. The PL was excited with either the 325 nm (above-gap) or 442 nm (belowgap) line of a HeCd laser.

Both samples exhibited characteristic 1.54 µm PL resulting from the intra-4f Er³⁺transition ⁴I_13/2→⁴I_15/2. The most striking feature of figure 1 is the large difference in PL intensity observed for the samples under above-gap excitation. The GaN: Er (SSMBE) sample exhibited a strong 1.54 µm PL which was nearly 80 times more intense than that observed from GaN: Er (MOMBE). On the contrary, under below-gap excitation both samples exhibited very similar 1.54 µm PL intensities. As shown previously [Reference Hömmerich, Seo, Thaik, Abernathy, MacKenzie and Zavada9], the weak PL observed under above-gap excitation from GaN: Er (MOMBE) can be explained by a significantly reduced excitation efficiency compared to below-gap excitation. Visible PL studies (see below) revealed, that for GaN: Er (MOMBE) the bandedge provides an efficient radiative combination channel reducing the excitation efficiency of intra-4f Er transitions. The excitation wavelength dependent PL study suggests that only weak electroluminescence can be expected from forward-biased GaN: Er (MOMBE) LED’s.

To further evaluate the GaN:Er samples for device applications, the temperature dependence of the integrated 1.54 µm PL intensities was measured as shown in Figure 2. Compared to Si: Er, GaAs: Er, and AlGaAs: Er [Reference Favennec, Haridon, Salvi, Moutonnet and Le Guillou10], both GaN: Er samples exhibited very stable 1.54 µm PL up to temperatures as high as ∼550 K. More information on the Er PL efficiency was obtained from temperature dependent lifetime studies. The total transition probability, i.e. the reciprocal of the experimental PL lifetime, is given as the sum of total radative decay rate, nonradiative decay rate through multiphonon relaxation, and nonradiative decay rate through energy transfer processes. It is assumed that at low temperature the nonradiative decay through either multiphonon relaxation and/or energy transfer is negligible small. Therefore, the lifetime at 15 K yields a good approximation for the radiative decay rate. Furthermore, assuming the radiative decay rate is temperature independent, any reduction in lifetime can be assigned to the onset of nonradiative decay. The luminescence transients of both GaN: Er samples were measured at 15 K, 300 K, and 520 K. It was observed that the decay curves were non-exponential which suggests the existence of multiple Er sites. The existence of multiple Er sites in GaN has been previously reported ¹¹. To describe the lifetime decay an average lifetime was used. For GaN: Er (SSMBE) it was observed that the low temperature (15 K) lifetime of 2.3 ms decreased to 1.9 ms at room temperature. At higher temperatures the lifetime continued to decrease and reached a value of 1.2 ms at 520 K. The decrease of the lifetime above 300K is most likely due to the onset of nonradiative decay. Compared to GaN: Er (SSMBE), the lifetime of GaN:Er (MOMBE) was significantly shorter and decreased slightly from 0.11 ms at 15 K to 0.10 ms at room temperature. At higher temperatures the lifetime was too short to be measured with our current setup. The room temperature luminescence efficiencies were estimated from the ratio of the low and room temperature lifetimes (τ_300K/τ_15K) to be ∼0.8 for GaN: Er (SSMBE) and ∼0.9 for GaN: Er (MOMBE), respectively. The high PL efficiencies indicate that the Er³⁺ excitation efficiency and the concentration of Er³⁺ ions limit the device performance of current infrared LED’s.

Figure 2: Comparison of the temperature dependence of the integrated Er³⁺ 1.54 µm PL for Er doped Si, GaAs, AlGaAs, and GaN (see also reference 10).

The visible PL spectra of the GaN: Er samples following optical excitation at 325 nm are shown in Figure 3. The GaN: Er (SSMBE) exhibited a weak bandedge PL at ∼369 nm (3.36 eV) and two “green” lines located at 537 nm and 558 nm. The green luminescence was assigned to the intra 4f Er³⁺ transitions ²H_11/2 → ⁴I_15/2 and ⁴S_3/2 → ⁴I_15/2 (Ref. 3). The GaN: Er (MOMBE) sample showed strong bandedge PL located at ∼381 nm (3.25 eV), however, no indication of green Er³⁺ luminescence was found. As discussed before, for GaN: Er (MOMBE) the bandedge provides an efficient radiative combination channel, which reduces the excitation efficiency for both infrared and visible Er³⁺ transitions. Figure 3b) shows the decay transients of the green Er³⁺ PL at different temperatures. The lifetime was found to be non-exponential at all temperatures and decreased with increasing temperature. The average lifetimes for the 558 nm line at 15 K and room temperature were determined to be 10.8 µs and 5.5 µs, respectively.

Figure 3: a) Visible PL spectra from GaN: Er (SSMBE) and GaN: Er MOMBE) at 300 K (λ_ex=325 nm). b) Decay transients of the visible PL at 558 nm from GaN: Er (SSMBE) at 15 K and 300 K (λ_ex=495 nm).

A more detailed study on the temperature dependence of the lifetime is depicted in Figure 4a). The thermalization of the ⁴S_3/2 and ²H_11/2 states leads to a common decay time τ (effective spontaneous emission probability), which can be described as:

Figure 4: a) Green PL lifetime of GaN: Er (SSMBE). The solid line describes the change in lifetime according to equation (1). b) Ratio of the PL intensities of the green lines as a function of temperature (see reference 3).

where τ_H and τ_S are the intrinsic radiative decay times of the ²H_11/2 → ⁴I_15/2 and ⁴S_3/2 → ⁴I_15/2 transitions, respectively. g_H and g_S are the electronic degeneracies (2J+1) of the ²H_11/2 and ⁴S_3/2 states and ΔE is their energy difference (ΔE=87meV) .

At low temperatures the ²H_11/2 state is not thermally populated and the experimental lifetime can be approximated as the intrinsic decay time of the ⁴S_3/2 → ⁴I_15/2 transition, i.e. τ_S=10.8 µs. It is assumed in this approximation that the low temperature decay time is purely radiative. The intrinsic lifetime of the ²H_11/2 → ⁴I_15/2 is not experimentally accessible, unless a careful analysis of absorption data is carried out, which is rather difficult for thin film materials. It is possible, however, to obtain a rough estimation of the intrinsic decay time τ_H from the temperature dependence of the luminescence intensity of the green Er³⁺ lines at 537 nm and 558 nm. Steckl and Birkhahn [Reference Steckl and Birkhahn3] reported that the intensity of the ⁴S_3/2 line decreased with increasing temperature, whereas the ²H_11/2 line had a maximum of intensity at around 300 K. Taking the experimental data from reference 3, the ratio of the 537 nm and 558 nm lines was calculated and is plotted in Fig 4b). Considering the thermal coupling of the involved states, the intensity ratio of both lines was fitted to an expression

with hω_H =2.309eV, hω_S=2.222eV, τ_S=10.8 µs and ΔE=87 meV. τ_H was taken as a fitting parameter and the best fit to the data yielded τ_H=0.75 µs. The fitting result shows that the radiative rate of the ²H_11/2 → ⁴I_15/2 transition is much larger than that of the ⁴S_3/2 → ⁴I_15/2 transition, consistent with published data on Er doped insulators. Using this set of parameters the temperature dependence of the luminescence lifetime was calculated according to equation (1) and is shown in Fig. 4a). The modeling reveals that the decrease of the luminescence lifetime with temperature is mainly due to an increased radiative decay rate arising from the fast thermalization of the ²H_11/2 → ⁴I_15/2 and ⁴S_3/2 → ⁴I_15/2 transitions. This preliminary analysis of the lifetime implies that non-radiative decay processes are small and therefore the green luminescence efficiency is high. To obtain further support for this conclusion it will be necessary to perform a more systematic study of the Er³⁺ visible PL lifetime for a series of samples with different Er concentrations.

Summary

In summary, we performed a comparison of the infrared and visible PL properties of GaN: Er (MOMBE) and GaN: Er (SSMBE). We observed that both samples exhibited intense 1.54 µm PL under below-gap excitation. With above-gap excitation GaN: Er (MOMBE) showed a greatly reduced IR PL intensity, whereas the 1.54 µm PL from GaN:Er (SSMBE) remained strong. Based on temperature dependent PL intensity and lifetime studies, it was concluded that the 1.54 µm luminescence efficiency is high (∼0.8-0.9). The factors limiting the performance of current IR LED’s are the Er excitation efficiency and the Er concentration. No visible PL arising from intra-4f Er transitions was found from GaN: Er (MOMBE). On the contrary, the GaN: Er (SSMBE) sample revealed green lines at 537 nm and 558 nm with an average lifetime of 5.5 µs at room temperature. The temperature dependence of the green lifetime was explained by the strong thermalization of the ²H_11/2 and ⁴S_3/2 excited states. Non-radiative decay does not seem to affect the green luminescence efficiency.