I. Introduction

III-nitride based materials have been intensively studied for applications such as visible light-emitting diodes (LEDs). While commercially viable blue and green LEDs can already be achieved using In_xGa_1-xN, red emission still remains a challenge. An InGaN-based red LED has been demonstrated, but suffered from problems such as a broad emission spectrum and current-dependent emission wavelength [1]. Alternatively, rare earth doped semiconductors exhibit narrow emission spectra, which is stable against temperature and the magnitude of injected current [2]. Eu-doped GaN (GaN:Eu) has received significant attention as a candidate for the active layer of a red GaN-based LED, which exhibits emission around 620nm due to intra-4f transitions from the ⁵D₀ to ⁷F₂ transition [3–7]. Substantial effort has gone into the enhancement of the emission intensity from GaN:Eu such as optimizing the growth conditions to improve crystal quality [8], co-doping with Mg or Si to facilitate the formation of new efficient luminescent centers [7,9,10], and improving the radiative transition probability by the modulating the photonic mode density [11]. However, the light output from GaN:Eu based devices remains insufficient for the practical use [12].

In our previous work, it was reported that eight unique Eu luminescent centers exist in GaN:Eu grown by organometallic vapor phase epitaxy (OMVPE), which have been labeled OMVPE 1 to 8 [13]. A specific luminescent center labeled OMVPE 7 was found to play an important role in the emission intensity under current injection [14]. Therefore, in order to obtain a higher light output from GaN:Eu red LEDs, it is necessary to increase the emission from OMVPE 7. Recently, it was shown that the photoluminescence (PL) intensity from the GaN:Eu grown on a sapphire substrate is larger than that grown on a free-standing GaN substrate, which was considered to be related to in-plane compressive strain within the GaN:Eu layer [15]. Since the magnitude of the strain can be controlled during crystal growth, the influence of strain on GaN:Eu may offer an alternative route to enhance its red luminescence. In this article, we investigate the effects of varying the in-plane compressive strain on the PL intensity from GaN:Eu by using AlGaN/AlN superlattices (SLs). It was revealed that compressive strain can be used to enhance the PL intensity from OMVPE 7, which was mainly due to an increase in its fractional percentage.

2. Experimental details

All samples were grown on (0001) sapphire substrates by OMVPE. Trimethylgallium, trimethylaluminum, and ammonia were used as sources for Ga, Al, and N, respectively. The Eu precursor was bis (n-propyl-tetramethylcyclopentadienyl) europium [(EuCp^pm₂)]. Oxygen was co-doped into GaN:Eu in order to enhance the PL intensity, and make emission spectrum sharper [10]. To introduce additional compressive strain, the undoped GaN buffer layer was followed by the growth of 10 pairs of AlN/AlGaN SLs. A 300 nm-thick Eu,O-codoped GaN (GaN:Eu,O) layer and a 20nm undoped GaN capping layer were grown on top of the SLs. The magnitude of the compressive strain was controlled by changing the Al composition of the AlGaN in the SLs. Three different samples were prepared; GaN:Eu,O without the SLs (hereafter referred to as w/o SLs) and GaN:Eu,O grown on the SLs with the Al compositions of 10 and 25% (with SLs: x = 0.1 and 0.25). The samples were characterized by the Raman spectroscopy, PL and time-resolved PL (TR-PL) using a He-Cd laser and a tunable dye laser as an excitation source. All experiments were carried out at 10 K by using the closed-cycle cryostat, in order to obtain sharp PL spectra. Under these conditions, the emission from different luminescent centers can be distinguished [10]. The emission spectra were recorded using a micro-PL system (Horiba: LabRAM HR-800). The signals of the PL decay were detected using a photon-counting system with a thermoelectrically cooled photomultiplier tube.

3. Results and discussion

Figure 1 shows the Raman spectra of the E₂^high mode from the GaN. The E₂^high mode contains information of the in-plane stress. The E₂^high mode shifted towards a higher wavenumber in the sample where the SLs have a higher Al composition. The in-plane compressive strain of each sample was estimated by $\Delta {\omega }_{\lambda }=2\left(a_{\lambda }-\frac{C_{13}}{C_{33}}{b}_{\lambda }\right)\epsilon$where $\Delta {\omega }_{\lambda }$ is the shift from GaN bulk, $a_{\lambda }$ and $b_{\lambda }$ are the deformation potential constants, $C_{13}$ and $C_{33}$ are the elastic constants. For GaN, $a_{\lambda }$, $b_{\lambda }$, $C_{13}$ and $C_{33}$ are −850 cm⁻¹, −920 cm⁻¹, 106 GPa and 398 GPa, respectively [16, 17]. From this equation, the strain was estimated to be −0.18, −0.36, and −0.72%, in the samples w/o SLs and with SLs: x = 0.1 and 0.25, respectively. A negative value for $\epsilon$ means that the strain in the GaN layer is compressive. Therefore, the in-plane compressive strain becomes larger with increased Al composition in the AlGaN layers. This is likely because the lattice constant of AlGaN is reduced with increased Al composition.

 

Fig. 1 Raman spectra of the E₂^high mode from the samples with different magnitudes of in-plane compressive strain.

Download Full Size | PDF

Figure 2(a) shows the PL spectra of each sample, which contain some of the peaks due to the ⁵D₀ – ⁷F₂ transition in the Eu ions. The emission at 1.9968 eV has been attributed to OMVPE 7, and the emission at 1.9941 eV has been associated with an overlap in the emission from OMVPE 4 and 7, which is hereafter referred to as OMVPE 4 + 7 [13,18].

 

Fig. 2 (a) PL spectra of samples with different magnitudes of in-plane compressive strain. (b) PL peak position energy as a function of in-plane compressive strain. (c) The integrated PL intensity dependence from OMVPE7 on the in-plane compressive strain.

Download Full Size | PDF

The dashed line in Fig. 2(a) denotes the emission wavelength of OMVPE 4 + 7 and OMVPE 7 in the sample w/o SLs. In Fig. 2(b), the dependence of the emission peak position on the magnitude of the in-plane compressive strain for OMVPE 4 + 7 and OMVPE 7 are compared for each sample. The Raman spectra in Fig. 1 only indicates that the GaN host material has suffered from in-plane compressive strain, however, this strain is also felt by the Eu ions, which can be seen from a red shift in the emission peak position of OMVPE 4 + 7 and OMVPE 7. It should be noted that the degree of red shift is larger for OMVPE 4 + 7 than for OMVPE 7. Figure 2(c) shows the dependence of the integrated PL intensity from OMVPE 7 on magnitude of the in-plane compressive strain. By inserting the SLs, the stress of the samples increased and the integrated PL intensity of OMVPE 7 was enhanced by 1.9 times. The results appear in contrary to the fact that the homo-epitaxial layer of GaN:Eu,O is expected to have the highest crystalline quality, and is expected to be the brightest. A similar relationship between strain and the emission from rare earth ions was also reported for GaN:Eu grown on the GaN bulk substrate or sapphire substrate [15], Er-doped Si grown on a Si or SiGe substrate [19], and Er-doped GaN grown on several different substrate materials [20]. However, none of these reports discussed the mechanism of emission enhancement. We predict that there are three primary mechanisms behind this emission enhancement: (1) the radiative transition probability of Eu ion, (2) energy transfer efficiency between the GaN host material and the Eu ion and (3) the number of the luminescent center. These three mechanisms will be investigated in the remainder of this report.

In order to explore the effect of strain on the radiative transition probability, TR-PL measurements were performed at 10K using a He-Cd laser, with a focus on the emission from OMVPE 7. The signals were fitted using a stretched exponential function, which allows for slight overlap of the emission from other luminescent centers. Using the fitting parameters, the PL emission lifetimes were found to be 228, 230 and 234 μs, in the samples w/o SLs, and with SLs: x = 0.1 and 0.25, respectively. These effective lifetimes are assumed to be radiative lifetimes, since all experiments were carried out at 10K. These results indicate that the radiative lifetime of OMVPE 7 increased slightly, despite the increased compressive strain. Therefore, it was concluded that the radiative transition probability does not play a role in the enhancement of the PL emission intensity from OMVPE 7. This tendency is consistent with a previous report of GaN:Eu grown on GaN bulk substrate and sapphire substrate [15].

To estimate the change of energy transfer efficiency through the excitation cross-section, the dependence of the PL emission intensity from OMVPE 7 on excitation power was measured, and is shown in Fig. 3. For each sample, the integrated emission intensities at each power were normalized to the respective intensity under an excitation power of 988 W/cm². The normalized integrated PL emission intensity of OMVPE 7 is always slightly higher in the sample w/o SLs than with SLs. This result indicates that the energy transfer efficiency is reduced by the addition of in-plane compressive strain. The excitation efficiency of the Eu ion is characterized by its excitation cross-section. In order to discuss the contribution of excitation efficiency on the enhancement of the PL intensity, a quantitative value for the excitation cross-section is necessary. The rate equation for the excitation and emission of the Eu³⁺ ions can be expressed as [21],

 

Fig. 3 The normalized PL intensity from OMVPE7 as a function of excitation power, for samples under varying magnitudes of in-plane compressive strain.

Download Full Size | PDF

To calculate the fractional percentage of each luminescent center, it is necessary to excite the Eu ions resonantly, from which information on the individual luminescent.

The fractional percentage of each luminescent center can be obtained from the PL intensity, the radiative lifetimes and excitation cross-sections under resonant excitation, using the following equation [22]:

 

Fig. 4 The fractional percentage of OMVPE 7 as a function of in-plane compressive strain.

Download Full Size | PDF

Using the model that the mechanism for the enhancement of the PL intensity is caused by an increase of the radiative transition probability, the energy transfer efficiency and the number of luminescent centers, an emission enhancement factor labeled ΔI_OMVPE7 can be estimated and compared with experiment. The Eu concentrations of the samples w/o SLs and with SLs: x = 0.25 were obtained using X-ray Fluorescence to be $3.4\times {10}^{19}$ cm⁻³ and $3.2\times {10}^{19}$ cm⁻³, respectively. Using these concentrations with the experimental results above, $\Delta I_{OMVPE\text{}7}$ was estimated to be ~1.5. On the other hand, a 1.9 fold enhancement of the integrated PL intensity was observed, as shown in Fig. 2(c). These results are comparable, therefore the prediction seems reasonable.

4. Conclusion

The integrated PL intensity from a Eu luminescent center in GaN:Eu, labelled OMVPE 7, was found to be enhanced by the addition of in-plane compressive strain. This enhancement was determined to be primarily a result of an increase in the number of Eu ions, which incorporated into GaN as this center. With the addition of compressive strain, the formation of OMVPE 7 more than doubled relative to the other Eu centers. Since OMVPE 7 is the most efficient Eu luminescent center under current injection, the manipulation of compressive strain is a promising method for enhancing the output intensity from GaN:Eu based red LEDs.