- Email: [email protected]
Eu3+ luminescence properties of Eu- and Mg-codoped AlGaN
Journal of Luminescence 166 (2015) 60–66
Contents lists available at ScienceDirect
Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin
Eu3 þ luminescence properties of Eu- and Mg-codoped AlGaN Masayoshi Kanemoto a, Hiroto Sekiguchi a,n, Keisuke Yamane a, Hiroshi Okada b,a, Akihiro Wakahara a,b a Department of Electrical and Electronics Information Engineering, Toyohashi University of Technology, 1-1 Hibarigaoka, Tempaku-cho, Toyohashi 441-8580, Aichi, Japan b Electronics-Inspired Interdisciplinary Research Institute (EIIRIS), Toyohashi University of Technology, 1-1 Hibarigaoka, Tempaku-cho, Toyohashi 441-8580, Aichi, Japan
art ic l e i nf o
a b s t r a c t
Article history: Received 14 February 2015 Received in revised form 22 April 2015 Accepted 23 April 2015 Available online 14 May 2015
We investigated the effect of Mg codoping on luminescence properties of AlGaN:Eu to improve emission through synergy effect between an increase in bandgap by AlGaN and the Mg codoping technique. The luminescence properties of AlGaN:(Eu, Mg) are strongly influenced by the Mg concentration and Al composition. Mg codoping in AlGaN was observed to contribute to increasing photoluminescence (PL) integrated intensity and to improve thermal quenching from 7.3% to 60% while the dominant optical site remained site B (622.3-nm peak) with low excitation cross section. The total concentration of optically activated Eu at 25 K was a constant at for either optical site, indicating that Mg codoping did not affect the formation of optical sites. The PL decay times at room temperature (RT) increased with Mg concentration because of suppression of the back-transfer process. For optimized Mg concentration, an increase in the Al composition contributed to the total activated Eu concentration and changed the dominant optical site from A (620.3-nm peak) to B. The activation energy Ea, which is the difference in energy between the 5D0 energy level and the trap level in the host material, was estimated from temperature dependence of PL decay time. The Ea for site A was larger than that for site B, suggesting that the back-transfer rate for site A was less than that for site B. & 2015 Elsevier B.V. All rights reserved.
Keywords: AlGaN Rare-earth doped semiconductor Eu Codoping
1. Introduction Electronic materials embedded with rare-earth (RE) ions are widely applied in optoelectronics devices, such as YAG:Nd solidstate laser, display phosphors and Er-doped optical-fiber amplifiers. The RE ion luminescence is based on the 4f–4f transition of innershell electrons shielded by outer shell electrons, showing a sharp line emission and thermal stability of the emission wavelength. However, miniaturization and integration of optical devices based on conventional RE-ion material are difficult because of the inability to excite RE ions through current injection. RE-doped semiconductors are potential candidates for realizing next-generation devices, such as optical devices with thermally stable emission wavelength and opto-spintronic devices. When a narrow-bandgap semiconductor (Si, GaAs, etc.) is selected as the host material, large thermal quenching and low photoluminescence (PL) intensity were observed at room temperature (RT) were observed [1]. Therefore, RE-doped nitride semiconductors with wide bandgaps have attracted attention for improving luminescence properties. Samples n
Corresponding author. E-mail address: [email protected] (H. Sekiguchi).
http://dx.doi.org/10.1016/j.jlumin.2015.04.036 0022-2313/& 2015 Elsevier B.V. All rights reserved.
have been prepared through ion implantation and annealing [2,3], in addition to crystal growth through molecular beam epitaxy (MBE) [4–8] and organometallic vapor phase epitaxy (OMVPE) [9]. RGB-color-integrated RE-doped GaN thin film electroluminescence (EL) devices, and light-emitting high electron-mobility transistors (HEMTs) with spatially selective Eu doping region have been fabricated [10,11]. Light-emitting diodes (LEDs) with Eu doped GaN (GaN:Eu) and Er doped InGaN active layers showing red and infrared emission have been demonstrated [12–14]. For further development of device performance, a fundamental technology for optically activating RE ions in a crystal is required. To develop such technology, selection of the host material and control of optical sites are key factors. It is well known that an increase in the bandgap of the host material contributes to suppressing the thermal quenching of PL intensity [1]. In fact, when the optical properties of Eu-, Tb-, Er-, and Tm-implanted AlGaN were evaluated, an increase in the Al composition resulted in an increase in the PL intensity from RE ions at RT owing to the suppression of the back-transfer process [2,15–17]. Therefore, the use of AlGaN with larger bandgap is promising to improve the luminescence efficiency. On the other hand, several emission peaks were observed in the Eu luminescence in GaN corresponding to 5 D0–7F2 electron transition, indicating the formation of several
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
61
The samples were grown on a GaN template by using NH3-source MBE. The group-III metals Ga (6N), Al (5N), and Eu (3N) as well as the Mg (6N) dopant were supplied from effusion cells. The beam equivalent pressure (BEP) of Ga, Al, and Eu were 6 10 7, 1.5 10 7, and 6 10 8 Torr, respectively. After cleaning with an organic solvent, the wafers were loaded into a chamber, and thermal cleaning was performed at 825 °C for 10 min. Subsequently, a GaN buffer layer was grown for 10 min, followed by the growth of an AlGaN:(Eu, Mg) layer with a NH3 flow of 3.0 sccm for 1 h. Al composition of AlGaN:(Eu, Mg) was determined to be 11% from the XRD pattern. The Eu concentration was estimated to be 2 1020 cm 3 by referring to the Eu concentration of Eu-doped GaN under the same growth condition of the quaternary alloys. Here, to investigate the dependence of optical properties on the Mg concentration (NMg), the Mg concentration was varied from 1.2 1017 to 8 1018 cm 3, which was estimated by using the secondary ion mass spectrometry (SIMS). In the PL measurement, a He–Cd laser (325 nm) and Xe lamp with band-pass filter (400 nm) were used as excitation sources. First, PL measurements of GaN:Eu and Al0.11Ga0.89N:Eu were performed to clarify the effect of the bandgap of the host material on the PL properties. Fig. 1 shows PL spectra at RT for GaN:Eu and
AlGaN:Eu. Sharp emission corresponding to 5D0–7F2 electron transition was observed around 620 nm for both samples. For GaN:Eu, three dominant peaks at 620.3 nm (shoulder peak), 622.3 nm and 633.8 nm were observed which are labeled peaks or sites A, B, and C for convenience, respectively. An additional other peak was observed at 625.5 nm. The difference in photon energy between peak B and this additional peak was approximately 12 meV, which corresponding to the value of energy localized to the Eu incorporation sites [19]. Therefore, the peak at 625.5 nm can be concluded to be a phonon replica of peak B, and it is labeled peak B. On the other hand, for AlGaN:Eu, a peak at 617.6 nm was observed in addition to peaks A, B, and C, additional peak is labeled peak or site D. To classify the optical sites according to the excitation process, the PL spectra of AlGaN:Eu sample under above-bandgap excitation (λex ¼325 nm) and below-bandgap excitation (λex ¼ 400 nm) were evaluated at RT. Peaks A, B, C, and D were observed under above-bandgap excitation, while only peak B was observed under below-bandgap excitation. This result suggests that the excitation process for sites A, C, and D is different from that for site B. The behaviors of optical properties corresponding to sites A and C were similar in this study. For example, the ratio of the PL intensity of peak A to that of peak C was almost constant among all samples despite of Mg concentration and Al composition. The excitation cross sections of sites A and C were almost the same. The origin of site D was not clear at this stage. However, the emission intensity corresponding to site D was sufficiently small compared with that of the dominant peak for either sample. Therefore, the optical property was discussed by focusing on the difference in behavior between sites A and B. The PL integrated intensity for AlGaN:Eu was 3.5 times as strong as that for GaN:Eu. To evaluate the effect of thermal quenching, the temperature dependence of PL was measured from 25 K to 300 K. The ratio of PL integrated intensity at 25 K to that at 300 K for AlGaN:Eu and GaN:Eu was 7.3% and 3.6%, respectively. Additionally, the 5D0-level lifetime at RT estimated from the timeresolved PL measurement for AlGaN:Eu and GaN:Eu were 30 and 17 μs, respectively. Therefore, AlGaN proved effective to improve the luminous efficiency. Next, the optical properties of AlGaN:(Eu, Mg) with different Mg concentrations were evaluated to investigate the synergy effect between Mg codoping and a wider bandgap. Fig. 2 shows the PL spectra at RT for AlGaN:(Eu, Mg) with different NMg and GaN:(Eu, Mg) with an optimized NMg of 3 1018 cm 3 as a
Fig. 1. PL spectra at RT for GaN:Eu and Al0.11Ga0.89N:Eu.
Fig. 2. PL spectra at RT for Al0.11Ga0.89N:(Eu, Mg) with different Mg concentrations and GaN:(Eu, Mg) with optimized an NMg of 3 1018 cm 3.
optical sites [18–22]. From combined excitation–emission spectroscopy (CEES), it was clarified that each optical site in GaN has a different excitation and emission efficiency [19]. Therefore, the control of optical sites can contribute to improve the optical properties. For Er-doped GaAs, oxygen codoping facilitated the formation of a specific local structure and resulted in the unification of the emission peak [23]. We developed an Mg-codoping technique for GaN:Eu grown by NH3-source MBE, which achieved the selective activation of one site and enhancement of PL intensity from Eu ions at RT [24,25]. The improvement of optical properties by Mg codoping has also been reported for OMVPEgrown Eu- and Mg-codoped GaN (GaN:(Eu, Mg)) and Eu-implanted Mg-doped GaN [26,27]. Based on these reports, in this study, we investigated the effect of Mg codoping on optical properties of AlGaN:Eu to improve emission through synergy effect between an increase in bandgap by AlGaN and the Mg codoping technique.
2. Effect of Mg codoping on optical properties of Eu-doped AlGaN
62
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
Fig. 3. (a) Arrhenius plots of PL integrated intensity of Al0.11Ga0.89N:(Eu, Mg) with different Mg concentrations. (b) Mg-concentration dependence of PL efficiency for GaN: (Eu, Mg) and Al0.11Ga0.89N:(Eu, Mg).
reference sample. Mg codoping into GaN:Eu induced a more than 10-fold increase in PL integrated intensity and selectively enhanced optical site A. For the case of AlGaN:(Eu, Mg), the PL integrated intensity increased on changing NMg and maximized at 3 1018 cm 3, which agrees with the optimized value for GaN:(Eu, Mg). The maximum PL integrated intensity was 3.4 times as large as that without Mg codoping, as the dominant optical site remained site B. At an NMg of 8 1018 cm 3, the PL intensity slightly decreased, indicating that excess NMg led to an increase in the nonradiative component. The PL integrated intensity of optimized GaN:(Eu, Mg) was approximately 1.5 times as large as that of optimized AlGaN:(Eu, Mg). Arrhenius plots of the PL integrated intensity of AlGaN:(Eu, Mg) with different NMg are shown in Fig. 3(a). In general, back transfer is suppressed at low temperature (LT), and the energy transfer is efficiently performed. Indeed, the difference in PL integrated intensity at 25 K was small among samples. Therefore, the ratio of the PL integrated intensity at 25 K to that at 300 K is an important indicator for evaluating the luminous efficiency, the value of which is defined as the PL efficiency. Fig. 3(b) shows the NMg dependence of PL efficiency estimated from the Arrhenius plot for AlGaN:(Eu, Mg) and GaN:(Eu, Mg). The PL efficiencies for AlGaN and GaN increased with increasing NMg from 7.3% to 60% and from 3.6% to 82%, respectively. PL efficiencies reached the maximum values at an NMg of 3 1018 cm 3 for both host materials. Excess Mg codoping of more than 3 1018 cm 3 caused a decrease in PL efficiency. The PL property from the Eu ion in AlGaN was determined by the product of optically activated Eu ions, excitation efficiency from the host to the Eu ion, and emission efficiency in the Eu ion because of a complex emission mechanism. To clarify the Mg codoping effect on each component for AlGaN:(Eu, Mg), the excitation power dependence of PL spectra was analyzed using rate equation described in Eq. (1), which yielded the total optically activated concentration of sites, the effective cross section corresponding to excitation efficiency, and the 5D0-level lifetime corresponding to emission efficiency in Eu ions
Fig. 4. Total optically activated concentrations of site x at 25 K and 300 K for Al0.11Ga0.89N:(Eu, Mg) and activation ratio as functions of Mg concentration.
dNx* N* = σϕx Nx − Nx* − x τx dt
(
)
(1)
where Nx* is the concentration of the excited Eu3 þ site x, Nx is the optically activated concentration of the Eu3 þ site x, sx is the effective cross section of site x, ϕex is the photon flux density, and τx is the 5D0-level lifetime of site x [28]. The PL spectra were deconvoluted to spectrally separate each site. Fig. 4 shows the total optically activated concentration of site x at 25 K and 300 K and activation ratio for AlGaN:(Eu, Mg) as a function of Mg concentration. It can be assumed that energy transfer is effectively performed at LT because no band-edge emission is observed in all samples and the nonradiative recombination and back-transfer process are expected to be sufficiently suppressed at LT. Therefore,
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
Nx at LT indicated the total site concentration corresponding to the number of active Eu sites with excitation path from AlGaN host material. Although NA and NB changed with changing NMg for GaN: (Eu, Mg) at 25 K [25], they did not change for AlGaN regardless of NMg. This suggested that Mg codoping did not influence the total optical site concentration in AlGaN. However, the effect of the back transfer process in the inner-shell relaxation process before the 5D0 level is occupied cannot be ignored with increasing the temperature, which resulted in the difference of the activation concentration between LT and RT. The ratio of Nx at 25 K to that at 300 K is defined as the activation ratio. While the activation ration of NB was constant regardless of Mg concentration, the activation of NA increased to approximately thrice as large as that without Mg at RT. Thus, one of the reasons for the Eu3 þ luminescence enhancement in AlGaN through Mg codoping is the reduction in back transfer for site A in the inner-shell relaxation process before the 5 D0 level is occupied. Next, the NMg dependence of excitation cross section corresponding to excitation efficiency was evaluated for AlGaN. The excitation cross section did not depend on NMg, for any of the optical sites (not shown), indicating that Mg codoping did not influence excitation efficiency. The excitation cross section for site A was larger than that for site B, which is in good agreement with other reports on GaN:Eu fabricated by ion implantation and GaN: (Eu, Mg) grown by MBE [20,25]. Finally, PL decay curves were evaluated to elucidate the effect of Mg concentration on the back-transfer process. In the timeresolved PL measurement, a fourth-harmonic-generation YAG pulsed laser was used for excitation and an electronic streak camera was used for evaluating temporal and spectral dispersion. As the peak separation of temporary and spectrally distributed results was difficult, PL decay curves at 25 K and RT of AlGaN:(Eu, Mg) with different NMg were measured by focusing on the dominant peak B, as shown in Fig. 5. The PL decay time was estimated
63
by fitting the stretched exponential decay function. The PL intensity at LT showed single-exponential decay, and the PL decay time was approximately 350 μs for all the samples. Thus, the difference of the PL decay time at RT simply indicates the difference of the nonradiative component resulting from the back-transfer process. The PL decay time at RT increased from 51 to 164 μs with increasing NMg, which suggested that Mg codoping was effective to reduce the nonradiative component resulting from the backtransfer process; that is, Mg codoping might contribute to decrease the native or unwanted defect in the AlGaN host as well as GaN [25]. Based on these results, the difference in the Mg codoping effect between GaN and AlGaN host materials is discussed. Mg codoping contributes to the suppression of nonradiative components caused by the back-transfer process in either host material. However, focusing on the optical sites, while the dominant optical site changed in GaN from site B with small excitation cross section to site A with large excitation cross section, no change in the dominant optical site was observed in AlGaN. The reason why the PL intensity of AlGaN:(Eu, Mg) was slightly smaller than that of GaN: (Eu, Mg) for optimized NMg, as shown in Fig. 2, is the difference in excitation efficiency. As described above, Eu ions of site B can be excited under below-bandgap excitation, which suggests that a shallow donor level is formed in site B; that is, the local structure of site B has a defect-complex structure around the Eu ion. In general, a higher growth temperature of AlGaN compared to that of GaN is required to obtain high-quality AlGaN. In this study, as AlGaN and GaN were grown at the same temperature, the crystal quality of AlGaN might be insufficient. In fact, when the PL properties at RT of AlGaN and GaN without Eu were evaluated, the near-band-edge intensity of AlGaN was one third that of GaN. If the crystal quality of AlGaN is improved, native or unwanted defects in AlGaN reduce, which might lead to a suppression in the formation of optical site B and an increase in the formation of
Fig. 5. Effect of Mg concentration on PL decay curves for site B measured at (a) 25 K and (b) RT.
64
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
optical site A. In the case of GaN, a switch in the dominant optical site was observed without a change in Eu concentration on increasing the growth temperature from 790 to 825 °C, as shown in Fig. 6. However, since the simple increase in growth temperature can influence Eu concentration and surface condition, an optimization of the growth condition considering the various views is required.
3. Effect of Al composition on optical properties of Eu- and Mg-codoped AlGaN The effect of Al composition on the optical properties of AlGaN: (Eu, Mg) was investigated. The BEP of Ga was 1.2 10 6 Torr, and the BEP of Al changed from 1.2 10 8 to 3.4 10 7 Torr. The Al composition (XAl) was varied from 0% to 27.6%, as estimated from XRD measurements. Eu and Mg concentrations were maintained at 4 1019 and 3 1018 cm 3, respectively. A 266-nm YAG pulsed laser was used to evaluate the samples under above-bandgap
Fig. 6. PL spectra at RT for GaN:Eu grown at 790 °C and 825 °C.
Fig. 7. PL spectra at RT for AlGaN:(Eu, Mg) with different Al compositions.
excitation. Fig. 7 shows PL spectra at RT for AlGaN:(Eu, Mg) with various XAl. Emission corresponding to site A was dominant for GaN. An increase in the XAl led to a decrease in the PL intensity of peak A and an increase in that of peak B. For XAl greater than 6.6%, peak B was split into two peaks. These two peaks can excite by both above and below bandgap excitations, suggesting that they were associated with site B. Therefore, their peaks are labeled peaks B1 and B2 for convenience. While the full widths of half maximum (FWHMs) of peaks A and B were 2.3 and 3.0 meV, the FWHMs of peaks B1 and B2 were increased to 3.8 and 4.1 meV, respectively. The increase in the FWHMs with increasing NAl could result from alloy disorder [15]. Next, the effect of XAl on the optical site was evaluated. Although an analysis using the steady-state solution of the rate equation (Eq. (1)) is adequate to clarify the number of optical sites, as discussed above, this method is not effective for measuring the effect of XAl on the optical site, because of the use of a pulsed laser with a pulse width of 5 ns. However, the excitation power intensity used here was very high, at approximately 100 μJ/pulse, which led to the excitation of most active Eu ions in AlGaN; that is, the PL integrated intensity became synonymous with the number of optically activated optical sites. Fig. 8 shows the ratio of PL integrated intensity of each site to that of whole sites (ηx) as a function of XAl. For GaN:(Eu, Mg), ηA and ηC were approximately 61% and 31%, respectively. While ηA and ηC monotonically decreased with increasing XAl, ηB dramatically increased from 8.1% to 93%. The dominant peak switched from site A to B for an XAl of 6.6%. Fig. 9 shows the PL integrated intensity and PL efficiency for each site and whole sites as a function of XAl. The whole PL integrated intensity at 25 K gradually increased twofold with increasing XAl. As described above, the nonradiative component was sufficiently suppressed at LT. Therefore, it is concluded that an increase in XAl leads to an increase in the number of activated optical sites. On the other hand, the PL integrated intensity at RT gradually decreased with increasing XAl below 16.6%, and increased with XAl until 27.6%, and the PL efficiency for whole sites decreased from 0.61 to 0.33 and saturated at an XAl of 27.6%. To clarify the reason for this change, the PL efficiency for each site was investigated. While the PL efficiency for site A changed from 0.61 to 0.76 with XAl changing from 0% to 6.6%, the PL efficiency for site B remained at approximately 0.33. Therefore, since the optical site switched from site A to B with increasing XAl, a change in PL efficiency can be interpreted as a result of weighting by the number of each optical site. In addition, the PL efficiency for site A was approximately 2 times that for site B. To understand this difference, the temperature dependence of the PL decay time for the dominant optical site was
Fig. 8. Ratio of PL integrated intensity of each site to that of whole sites (ηx) as a function of Al composition.
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
65
transfer process. The obtained values are summarized in Table 1. The activation energy of site A was 2.5–3 times that of site B, indicating that the energy back transfer for site A was suppressed compared with that for site B. This resulted in high PL efficiency for site A. In addition, an increase in XAl tended to increase the activation energy for both sites, which could be caused by the trap-level-related Eu ions in AlGaN shifting to higher energies with increasing bandgap. The effect of XAl is summarized as follows: the dominant site B of Eu- and Mg-codoped AlGaN with a higher back-transfer rate led to a decrease in the PL efficiency, but increases in optical sites and activation energy with increasing XAl contributed to a slightly higher PL intensity compared with that of GaN:(Eu, Mg), indicating that a further increase in XAl can improve the optical properties. As described in the previous section, if the crystal quality of AlGaN host materials is improved by an increase in the growth temperature or application of annealing process, the dominant site becomes site A for AlGaN, and the emission efficiency can be dramatically improved.
4. Summary Fig. 9. PL integrated intensities at 25 K and 300 K (above) and PL efficiencies for sites A and B and whole sites (below) as functions of Al composition.
Mg-concentration and Al-composition dependences of the optical properties of AlGaN:(Eu, Mg) were investigated to clarify the synergy effect between Mg codoping and a wider bandgap. Mg codoping into AlGaN contributed to increase the PL intensity and to improve the PL efficiency from 7.3% to 60% while the dominant optical site remained site B with low excitation cross section. Optically activated Eu concentration at LT for sites A and B remained constant, suggesting that Mg codoping did not influence the incorporated site. The 5D0-level lifetime at RT increased with Mg concentration owing to the suppression of the back-transfer process. A change in Al composition at the optimized Mg concentration resulted in the dominant switching from sites A to B and increased the total activated Eu concentration. The activation energy, which represented the difference of energy between 5D0 level and trap level, for site A was larger than that for site B, suggesting that the energy back transfer for site A was suppressed compared with that for site B.
Acknowledgment
Fig. 10. Arrhenius plots of PL decay time of the dominant site for AlGaN:(Eu, Mg).
Table 1 Activation energy of AlGaN:(Eu, Mg) with different Al compositions. Al composition (%) 27.6 Site Activation energy (meV)
0
6.6
6.6
A 161
A 198
B 63
16.6 B 68
B 113
investigated. The PL decay time of site A for XAl values of 0% and 6.6% and those of site B for XAl greater than 6.6% were evaluated. The Arrhenius plots of PL decay time of each site for AlGaN:(Eu, Mg) are shown in Fig. 10. In the phonon-assisted back-transfer model suggested in Ref. [28], the temperature dependence of PL decay time can be expressed as
⎛ Ea ⎞ 1 1 1 exp ⎜ − ⎟ = + ⎝ kT ⎠ τx (T ) τ0 τBT
(2)
where τ0 is the decay time at LT, τBT is a fitting parameter, and Ea is the activation energy, which is the difference in energy between the 5D0 level and the trap level in the host related to the back-
This work was partly supported by a Grant-in-Aid Scientific Research (C) #26420271 from the Ministry of Education, Culture, Sports, Science and Technology in Japan and the Toyoaki Scholarship Foundation.
References [1] P.N. Favennec, H. L'Haridon, M. Salvi, D. Moutonnet, Y.L. Guillou, Electron. Lett. 25 (1989) 718. [2] A. Wakahara, Opt. Mater. 28 (2006) 731. [3] K. Wang, R.W. Martin, K.P. O'Donnell, V. Katchkanov, E. Nogales, K. Lorenz, E. Alves, S. Ruffenach, O. Briot, Appl. Phys. Lett. 87 (2005) 112107. [4] S. Morishima, T. Maruyama, K. Akimoto, J. Cryst. Growth 209 (2000) 378. [5] E.E. Nyein, U. Hömmerich, C. Munasinghe, A.J. Steckl, J.M. Zavada, Mater. Res. Soc. Symp. Proc. 866 (V3) (2005) 5. [6] R. Wang, A.J. Steckl, J. Cryst. Growth 312 (2010) 680. [7] M. Zhong, A.J. Steckl, Appl. Phys. Express 3 (2010) 121003. [8] J.-H. Park, A. Wakahara, H. Okada, H. Sekiguchi, A. Tiwari, Y.-T. Kim, J. Song, J.-H. Lee, J. Jhin, Jpn. J. Appl. Phys. 50 (2011) 031003. [9] M. Pan, A.J. Steckl, Appl. Phys. Lett. 83 (2003) 9. [10] Y.Q. Wang, A.J. Steckl, Appl. Phys. Lett. 82 (2002) 502. [11] H. Okada, K. Takemoto, F. Oikawa, Y. Furukawa, A. Wakahara, S. Sato, T. Ohshima, Phys. Status Solidi C 6 (2009) S631. [12] A. Nishikawa, T. Kawasaki, N. Furukawa, Y. Terai, Y. Fujiwara, Appl. Phys. Express 2 (2009) 071004.
66
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
[13] H. Sekiguchi, Y. Takagi, T. Otani, R. Matsumura, H. Okada, A. Wakahara, Jpn. J. Appl. Phys. 52 (2013). [14] R. Dahal, C. Ugolini, J.Y. Lin, H.X. Jiang, J.M. Zavada, Appl. Phys. Lett. 97 (2010) 141109. [15] K. Wang, K.P. O'Donnell, B. Hourahine, R.W. Martin, I.M. Watson, K. Lorenz, E. Alves, Phys. Rev. B 80 (2009) 125206. [16] U. Hömmerich, E.E. Nyein, D.S. Lee, J. Heikenfeld, A.J. Steckl, J.M. Zavada, Mater. Sci. Eng. B 105 (2003) 91. [17] Y. Nakanishi, A. Wakahara, H. Okada, A. Yoshida, T. Ohshima, H. Itoh, T. Shibata, M. Tanaka, Phys. Status Solidi C 0 (2003) 2623. [18] E.E. Nyein, U. Hommerich, J. Heikenfeld, D.S. Lee, A.J. Steckl, J.M. Zavada, Appl. Phys. Lett. 82 (2003) 1655. [19] A. Fleischman, C. Munasinghe, A.J. Steckl, A. Wakahara, J. Zavada, V. Dierolf, Appl. Phys. B 97 (2009) 607. [20] L. Bodiou, A. Oussif, A. Braud, J.-L. Doualan, R. Moncorge, K. Lorenz, E. Alves, Opt. Mater. 28 (2006) 780.
[21] I.S. Roqan, K.P. O'Donnell, R.W. Martin, P.R. Edwards, S.F. Song, A. Vantomme, K. Lorenz, E. Alves, M. Bockowski, Phys. Rev. B 81 (2010) 085209. [22] N. Woodward, J. Poplawsky, B. Mitchell, A. Nishikawa, Y. Fujiwara, V. Dierolf, Appl. Phys. Lett. 98 (2011) 011102. [23] K. Takahei, A. Taguchi, J. Appl. Phys. 74 (1993) 1979. [24] Y. Takagi, T. Suwa, H. Sekiguchi, H. Okada, A. Wakahara, Appl. Phys. Lett. 99 (2011) 171905. [25] H. Sekiguchi, Y. Takagi, T. Otani, H. Okada, A. Wakahara, J. Appl. Phys. 113 (2013) 013105. [26] D.-G. Lee, A. Nishikawa, Y. Terai, Y. Fujiwara, Appl. Phys. Lett. 100 (2012) 171904. [27] J.K. Mishra, T. Langer, U. Rossow, S. Shvakov, A. Wieck, A. Hangleiter, Appl. Phys. Lett. 102 (2013) 061115. [28] F. Priolo, G. Dranzo, S. Coffa, A. Carnera, Phys. Rev. B 57 (1998) 4443.
Contents lists available at ScienceDirect
Journal of Luminescence journal homepage: www.elsevier.com/locate/jlumin
Eu3 þ luminescence properties of Eu- and Mg-codoped AlGaN Masayoshi Kanemoto a, Hiroto Sekiguchi a,n, Keisuke Yamane a, Hiroshi Okada b,a, Akihiro Wakahara a,b a Department of Electrical and Electronics Information Engineering, Toyohashi University of Technology, 1-1 Hibarigaoka, Tempaku-cho, Toyohashi 441-8580, Aichi, Japan b Electronics-Inspired Interdisciplinary Research Institute (EIIRIS), Toyohashi University of Technology, 1-1 Hibarigaoka, Tempaku-cho, Toyohashi 441-8580, Aichi, Japan
art ic l e i nf o
a b s t r a c t
Article history: Received 14 February 2015 Received in revised form 22 April 2015 Accepted 23 April 2015 Available online 14 May 2015
We investigated the effect of Mg codoping on luminescence properties of AlGaN:Eu to improve emission through synergy effect between an increase in bandgap by AlGaN and the Mg codoping technique. The luminescence properties of AlGaN:(Eu, Mg) are strongly influenced by the Mg concentration and Al composition. Mg codoping in AlGaN was observed to contribute to increasing photoluminescence (PL) integrated intensity and to improve thermal quenching from 7.3% to 60% while the dominant optical site remained site B (622.3-nm peak) with low excitation cross section. The total concentration of optically activated Eu at 25 K was a constant at for either optical site, indicating that Mg codoping did not affect the formation of optical sites. The PL decay times at room temperature (RT) increased with Mg concentration because of suppression of the back-transfer process. For optimized Mg concentration, an increase in the Al composition contributed to the total activated Eu concentration and changed the dominant optical site from A (620.3-nm peak) to B. The activation energy Ea, which is the difference in energy between the 5D0 energy level and the trap level in the host material, was estimated from temperature dependence of PL decay time. The Ea for site A was larger than that for site B, suggesting that the back-transfer rate for site A was less than that for site B. & 2015 Elsevier B.V. All rights reserved.
Keywords: AlGaN Rare-earth doped semiconductor Eu Codoping
1. Introduction Electronic materials embedded with rare-earth (RE) ions are widely applied in optoelectronics devices, such as YAG:Nd solidstate laser, display phosphors and Er-doped optical-fiber amplifiers. The RE ion luminescence is based on the 4f–4f transition of innershell electrons shielded by outer shell electrons, showing a sharp line emission and thermal stability of the emission wavelength. However, miniaturization and integration of optical devices based on conventional RE-ion material are difficult because of the inability to excite RE ions through current injection. RE-doped semiconductors are potential candidates for realizing next-generation devices, such as optical devices with thermally stable emission wavelength and opto-spintronic devices. When a narrow-bandgap semiconductor (Si, GaAs, etc.) is selected as the host material, large thermal quenching and low photoluminescence (PL) intensity were observed at room temperature (RT) were observed [1]. Therefore, RE-doped nitride semiconductors with wide bandgaps have attracted attention for improving luminescence properties. Samples n
Corresponding author. E-mail address: [email protected] (H. Sekiguchi).
http://dx.doi.org/10.1016/j.jlumin.2015.04.036 0022-2313/& 2015 Elsevier B.V. All rights reserved.
have been prepared through ion implantation and annealing [2,3], in addition to crystal growth through molecular beam epitaxy (MBE) [4–8] and organometallic vapor phase epitaxy (OMVPE) [9]. RGB-color-integrated RE-doped GaN thin film electroluminescence (EL) devices, and light-emitting high electron-mobility transistors (HEMTs) with spatially selective Eu doping region have been fabricated [10,11]. Light-emitting diodes (LEDs) with Eu doped GaN (GaN:Eu) and Er doped InGaN active layers showing red and infrared emission have been demonstrated [12–14]. For further development of device performance, a fundamental technology for optically activating RE ions in a crystal is required. To develop such technology, selection of the host material and control of optical sites are key factors. It is well known that an increase in the bandgap of the host material contributes to suppressing the thermal quenching of PL intensity [1]. In fact, when the optical properties of Eu-, Tb-, Er-, and Tm-implanted AlGaN were evaluated, an increase in the Al composition resulted in an increase in the PL intensity from RE ions at RT owing to the suppression of the back-transfer process [2,15–17]. Therefore, the use of AlGaN with larger bandgap is promising to improve the luminescence efficiency. On the other hand, several emission peaks were observed in the Eu luminescence in GaN corresponding to 5 D0–7F2 electron transition, indicating the formation of several
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
61
The samples were grown on a GaN template by using NH3-source MBE. The group-III metals Ga (6N), Al (5N), and Eu (3N) as well as the Mg (6N) dopant were supplied from effusion cells. The beam equivalent pressure (BEP) of Ga, Al, and Eu were 6 10 7, 1.5 10 7, and 6 10 8 Torr, respectively. After cleaning with an organic solvent, the wafers were loaded into a chamber, and thermal cleaning was performed at 825 °C for 10 min. Subsequently, a GaN buffer layer was grown for 10 min, followed by the growth of an AlGaN:(Eu, Mg) layer with a NH3 flow of 3.0 sccm for 1 h. Al composition of AlGaN:(Eu, Mg) was determined to be 11% from the XRD pattern. The Eu concentration was estimated to be 2 1020 cm 3 by referring to the Eu concentration of Eu-doped GaN under the same growth condition of the quaternary alloys. Here, to investigate the dependence of optical properties on the Mg concentration (NMg), the Mg concentration was varied from 1.2 1017 to 8 1018 cm 3, which was estimated by using the secondary ion mass spectrometry (SIMS). In the PL measurement, a He–Cd laser (325 nm) and Xe lamp with band-pass filter (400 nm) were used as excitation sources. First, PL measurements of GaN:Eu and Al0.11Ga0.89N:Eu were performed to clarify the effect of the bandgap of the host material on the PL properties. Fig. 1 shows PL spectra at RT for GaN:Eu and
AlGaN:Eu. Sharp emission corresponding to 5D0–7F2 electron transition was observed around 620 nm for both samples. For GaN:Eu, three dominant peaks at 620.3 nm (shoulder peak), 622.3 nm and 633.8 nm were observed which are labeled peaks or sites A, B, and C for convenience, respectively. An additional other peak was observed at 625.5 nm. The difference in photon energy between peak B and this additional peak was approximately 12 meV, which corresponding to the value of energy localized to the Eu incorporation sites [19]. Therefore, the peak at 625.5 nm can be concluded to be a phonon replica of peak B, and it is labeled peak B. On the other hand, for AlGaN:Eu, a peak at 617.6 nm was observed in addition to peaks A, B, and C, additional peak is labeled peak or site D. To classify the optical sites according to the excitation process, the PL spectra of AlGaN:Eu sample under above-bandgap excitation (λex ¼325 nm) and below-bandgap excitation (λex ¼ 400 nm) were evaluated at RT. Peaks A, B, C, and D were observed under above-bandgap excitation, while only peak B was observed under below-bandgap excitation. This result suggests that the excitation process for sites A, C, and D is different from that for site B. The behaviors of optical properties corresponding to sites A and C were similar in this study. For example, the ratio of the PL intensity of peak A to that of peak C was almost constant among all samples despite of Mg concentration and Al composition. The excitation cross sections of sites A and C were almost the same. The origin of site D was not clear at this stage. However, the emission intensity corresponding to site D was sufficiently small compared with that of the dominant peak for either sample. Therefore, the optical property was discussed by focusing on the difference in behavior between sites A and B. The PL integrated intensity for AlGaN:Eu was 3.5 times as strong as that for GaN:Eu. To evaluate the effect of thermal quenching, the temperature dependence of PL was measured from 25 K to 300 K. The ratio of PL integrated intensity at 25 K to that at 300 K for AlGaN:Eu and GaN:Eu was 7.3% and 3.6%, respectively. Additionally, the 5D0-level lifetime at RT estimated from the timeresolved PL measurement for AlGaN:Eu and GaN:Eu were 30 and 17 μs, respectively. Therefore, AlGaN proved effective to improve the luminous efficiency. Next, the optical properties of AlGaN:(Eu, Mg) with different Mg concentrations were evaluated to investigate the synergy effect between Mg codoping and a wider bandgap. Fig. 2 shows the PL spectra at RT for AlGaN:(Eu, Mg) with different NMg and GaN:(Eu, Mg) with an optimized NMg of 3 1018 cm 3 as a
Fig. 1. PL spectra at RT for GaN:Eu and Al0.11Ga0.89N:Eu.
Fig. 2. PL spectra at RT for Al0.11Ga0.89N:(Eu, Mg) with different Mg concentrations and GaN:(Eu, Mg) with optimized an NMg of 3 1018 cm 3.
optical sites [18–22]. From combined excitation–emission spectroscopy (CEES), it was clarified that each optical site in GaN has a different excitation and emission efficiency [19]. Therefore, the control of optical sites can contribute to improve the optical properties. For Er-doped GaAs, oxygen codoping facilitated the formation of a specific local structure and resulted in the unification of the emission peak [23]. We developed an Mg-codoping technique for GaN:Eu grown by NH3-source MBE, which achieved the selective activation of one site and enhancement of PL intensity from Eu ions at RT [24,25]. The improvement of optical properties by Mg codoping has also been reported for OMVPEgrown Eu- and Mg-codoped GaN (GaN:(Eu, Mg)) and Eu-implanted Mg-doped GaN [26,27]. Based on these reports, in this study, we investigated the effect of Mg codoping on optical properties of AlGaN:Eu to improve emission through synergy effect between an increase in bandgap by AlGaN and the Mg codoping technique.
2. Effect of Mg codoping on optical properties of Eu-doped AlGaN
62
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
Fig. 3. (a) Arrhenius plots of PL integrated intensity of Al0.11Ga0.89N:(Eu, Mg) with different Mg concentrations. (b) Mg-concentration dependence of PL efficiency for GaN: (Eu, Mg) and Al0.11Ga0.89N:(Eu, Mg).
reference sample. Mg codoping into GaN:Eu induced a more than 10-fold increase in PL integrated intensity and selectively enhanced optical site A. For the case of AlGaN:(Eu, Mg), the PL integrated intensity increased on changing NMg and maximized at 3 1018 cm 3, which agrees with the optimized value for GaN:(Eu, Mg). The maximum PL integrated intensity was 3.4 times as large as that without Mg codoping, as the dominant optical site remained site B. At an NMg of 8 1018 cm 3, the PL intensity slightly decreased, indicating that excess NMg led to an increase in the nonradiative component. The PL integrated intensity of optimized GaN:(Eu, Mg) was approximately 1.5 times as large as that of optimized AlGaN:(Eu, Mg). Arrhenius plots of the PL integrated intensity of AlGaN:(Eu, Mg) with different NMg are shown in Fig. 3(a). In general, back transfer is suppressed at low temperature (LT), and the energy transfer is efficiently performed. Indeed, the difference in PL integrated intensity at 25 K was small among samples. Therefore, the ratio of the PL integrated intensity at 25 K to that at 300 K is an important indicator for evaluating the luminous efficiency, the value of which is defined as the PL efficiency. Fig. 3(b) shows the NMg dependence of PL efficiency estimated from the Arrhenius plot for AlGaN:(Eu, Mg) and GaN:(Eu, Mg). The PL efficiencies for AlGaN and GaN increased with increasing NMg from 7.3% to 60% and from 3.6% to 82%, respectively. PL efficiencies reached the maximum values at an NMg of 3 1018 cm 3 for both host materials. Excess Mg codoping of more than 3 1018 cm 3 caused a decrease in PL efficiency. The PL property from the Eu ion in AlGaN was determined by the product of optically activated Eu ions, excitation efficiency from the host to the Eu ion, and emission efficiency in the Eu ion because of a complex emission mechanism. To clarify the Mg codoping effect on each component for AlGaN:(Eu, Mg), the excitation power dependence of PL spectra was analyzed using rate equation described in Eq. (1), which yielded the total optically activated concentration of sites, the effective cross section corresponding to excitation efficiency, and the 5D0-level lifetime corresponding to emission efficiency in Eu ions
Fig. 4. Total optically activated concentrations of site x at 25 K and 300 K for Al0.11Ga0.89N:(Eu, Mg) and activation ratio as functions of Mg concentration.
dNx* N* = σϕx Nx − Nx* − x τx dt
(
)
(1)
where Nx* is the concentration of the excited Eu3 þ site x, Nx is the optically activated concentration of the Eu3 þ site x, sx is the effective cross section of site x, ϕex is the photon flux density, and τx is the 5D0-level lifetime of site x [28]. The PL spectra were deconvoluted to spectrally separate each site. Fig. 4 shows the total optically activated concentration of site x at 25 K and 300 K and activation ratio for AlGaN:(Eu, Mg) as a function of Mg concentration. It can be assumed that energy transfer is effectively performed at LT because no band-edge emission is observed in all samples and the nonradiative recombination and back-transfer process are expected to be sufficiently suppressed at LT. Therefore,
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
Nx at LT indicated the total site concentration corresponding to the number of active Eu sites with excitation path from AlGaN host material. Although NA and NB changed with changing NMg for GaN: (Eu, Mg) at 25 K [25], they did not change for AlGaN regardless of NMg. This suggested that Mg codoping did not influence the total optical site concentration in AlGaN. However, the effect of the back transfer process in the inner-shell relaxation process before the 5D0 level is occupied cannot be ignored with increasing the temperature, which resulted in the difference of the activation concentration between LT and RT. The ratio of Nx at 25 K to that at 300 K is defined as the activation ratio. While the activation ration of NB was constant regardless of Mg concentration, the activation of NA increased to approximately thrice as large as that without Mg at RT. Thus, one of the reasons for the Eu3 þ luminescence enhancement in AlGaN through Mg codoping is the reduction in back transfer for site A in the inner-shell relaxation process before the 5 D0 level is occupied. Next, the NMg dependence of excitation cross section corresponding to excitation efficiency was evaluated for AlGaN. The excitation cross section did not depend on NMg, for any of the optical sites (not shown), indicating that Mg codoping did not influence excitation efficiency. The excitation cross section for site A was larger than that for site B, which is in good agreement with other reports on GaN:Eu fabricated by ion implantation and GaN: (Eu, Mg) grown by MBE [20,25]. Finally, PL decay curves were evaluated to elucidate the effect of Mg concentration on the back-transfer process. In the timeresolved PL measurement, a fourth-harmonic-generation YAG pulsed laser was used for excitation and an electronic streak camera was used for evaluating temporal and spectral dispersion. As the peak separation of temporary and spectrally distributed results was difficult, PL decay curves at 25 K and RT of AlGaN:(Eu, Mg) with different NMg were measured by focusing on the dominant peak B, as shown in Fig. 5. The PL decay time was estimated
63
by fitting the stretched exponential decay function. The PL intensity at LT showed single-exponential decay, and the PL decay time was approximately 350 μs for all the samples. Thus, the difference of the PL decay time at RT simply indicates the difference of the nonradiative component resulting from the back-transfer process. The PL decay time at RT increased from 51 to 164 μs with increasing NMg, which suggested that Mg codoping was effective to reduce the nonradiative component resulting from the backtransfer process; that is, Mg codoping might contribute to decrease the native or unwanted defect in the AlGaN host as well as GaN [25]. Based on these results, the difference in the Mg codoping effect between GaN and AlGaN host materials is discussed. Mg codoping contributes to the suppression of nonradiative components caused by the back-transfer process in either host material. However, focusing on the optical sites, while the dominant optical site changed in GaN from site B with small excitation cross section to site A with large excitation cross section, no change in the dominant optical site was observed in AlGaN. The reason why the PL intensity of AlGaN:(Eu, Mg) was slightly smaller than that of GaN: (Eu, Mg) for optimized NMg, as shown in Fig. 2, is the difference in excitation efficiency. As described above, Eu ions of site B can be excited under below-bandgap excitation, which suggests that a shallow donor level is formed in site B; that is, the local structure of site B has a defect-complex structure around the Eu ion. In general, a higher growth temperature of AlGaN compared to that of GaN is required to obtain high-quality AlGaN. In this study, as AlGaN and GaN were grown at the same temperature, the crystal quality of AlGaN might be insufficient. In fact, when the PL properties at RT of AlGaN and GaN without Eu were evaluated, the near-band-edge intensity of AlGaN was one third that of GaN. If the crystal quality of AlGaN is improved, native or unwanted defects in AlGaN reduce, which might lead to a suppression in the formation of optical site B and an increase in the formation of
Fig. 5. Effect of Mg concentration on PL decay curves for site B measured at (a) 25 K and (b) RT.
64
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
optical site A. In the case of GaN, a switch in the dominant optical site was observed without a change in Eu concentration on increasing the growth temperature from 790 to 825 °C, as shown in Fig. 6. However, since the simple increase in growth temperature can influence Eu concentration and surface condition, an optimization of the growth condition considering the various views is required.
3. Effect of Al composition on optical properties of Eu- and Mg-codoped AlGaN The effect of Al composition on the optical properties of AlGaN: (Eu, Mg) was investigated. The BEP of Ga was 1.2 10 6 Torr, and the BEP of Al changed from 1.2 10 8 to 3.4 10 7 Torr. The Al composition (XAl) was varied from 0% to 27.6%, as estimated from XRD measurements. Eu and Mg concentrations were maintained at 4 1019 and 3 1018 cm 3, respectively. A 266-nm YAG pulsed laser was used to evaluate the samples under above-bandgap
Fig. 6. PL spectra at RT for GaN:Eu grown at 790 °C and 825 °C.
Fig. 7. PL spectra at RT for AlGaN:(Eu, Mg) with different Al compositions.
excitation. Fig. 7 shows PL spectra at RT for AlGaN:(Eu, Mg) with various XAl. Emission corresponding to site A was dominant for GaN. An increase in the XAl led to a decrease in the PL intensity of peak A and an increase in that of peak B. For XAl greater than 6.6%, peak B was split into two peaks. These two peaks can excite by both above and below bandgap excitations, suggesting that they were associated with site B. Therefore, their peaks are labeled peaks B1 and B2 for convenience. While the full widths of half maximum (FWHMs) of peaks A and B were 2.3 and 3.0 meV, the FWHMs of peaks B1 and B2 were increased to 3.8 and 4.1 meV, respectively. The increase in the FWHMs with increasing NAl could result from alloy disorder [15]. Next, the effect of XAl on the optical site was evaluated. Although an analysis using the steady-state solution of the rate equation (Eq. (1)) is adequate to clarify the number of optical sites, as discussed above, this method is not effective for measuring the effect of XAl on the optical site, because of the use of a pulsed laser with a pulse width of 5 ns. However, the excitation power intensity used here was very high, at approximately 100 μJ/pulse, which led to the excitation of most active Eu ions in AlGaN; that is, the PL integrated intensity became synonymous with the number of optically activated optical sites. Fig. 8 shows the ratio of PL integrated intensity of each site to that of whole sites (ηx) as a function of XAl. For GaN:(Eu, Mg), ηA and ηC were approximately 61% and 31%, respectively. While ηA and ηC monotonically decreased with increasing XAl, ηB dramatically increased from 8.1% to 93%. The dominant peak switched from site A to B for an XAl of 6.6%. Fig. 9 shows the PL integrated intensity and PL efficiency for each site and whole sites as a function of XAl. The whole PL integrated intensity at 25 K gradually increased twofold with increasing XAl. As described above, the nonradiative component was sufficiently suppressed at LT. Therefore, it is concluded that an increase in XAl leads to an increase in the number of activated optical sites. On the other hand, the PL integrated intensity at RT gradually decreased with increasing XAl below 16.6%, and increased with XAl until 27.6%, and the PL efficiency for whole sites decreased from 0.61 to 0.33 and saturated at an XAl of 27.6%. To clarify the reason for this change, the PL efficiency for each site was investigated. While the PL efficiency for site A changed from 0.61 to 0.76 with XAl changing from 0% to 6.6%, the PL efficiency for site B remained at approximately 0.33. Therefore, since the optical site switched from site A to B with increasing XAl, a change in PL efficiency can be interpreted as a result of weighting by the number of each optical site. In addition, the PL efficiency for site A was approximately 2 times that for site B. To understand this difference, the temperature dependence of the PL decay time for the dominant optical site was
Fig. 8. Ratio of PL integrated intensity of each site to that of whole sites (ηx) as a function of Al composition.
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
65
transfer process. The obtained values are summarized in Table 1. The activation energy of site A was 2.5–3 times that of site B, indicating that the energy back transfer for site A was suppressed compared with that for site B. This resulted in high PL efficiency for site A. In addition, an increase in XAl tended to increase the activation energy for both sites, which could be caused by the trap-level-related Eu ions in AlGaN shifting to higher energies with increasing bandgap. The effect of XAl is summarized as follows: the dominant site B of Eu- and Mg-codoped AlGaN with a higher back-transfer rate led to a decrease in the PL efficiency, but increases in optical sites and activation energy with increasing XAl contributed to a slightly higher PL intensity compared with that of GaN:(Eu, Mg), indicating that a further increase in XAl can improve the optical properties. As described in the previous section, if the crystal quality of AlGaN host materials is improved by an increase in the growth temperature or application of annealing process, the dominant site becomes site A for AlGaN, and the emission efficiency can be dramatically improved.
4. Summary Fig. 9. PL integrated intensities at 25 K and 300 K (above) and PL efficiencies for sites A and B and whole sites (below) as functions of Al composition.
Mg-concentration and Al-composition dependences of the optical properties of AlGaN:(Eu, Mg) were investigated to clarify the synergy effect between Mg codoping and a wider bandgap. Mg codoping into AlGaN contributed to increase the PL intensity and to improve the PL efficiency from 7.3% to 60% while the dominant optical site remained site B with low excitation cross section. Optically activated Eu concentration at LT for sites A and B remained constant, suggesting that Mg codoping did not influence the incorporated site. The 5D0-level lifetime at RT increased with Mg concentration owing to the suppression of the back-transfer process. A change in Al composition at the optimized Mg concentration resulted in the dominant switching from sites A to B and increased the total activated Eu concentration. The activation energy, which represented the difference of energy between 5D0 level and trap level, for site A was larger than that for site B, suggesting that the energy back transfer for site A was suppressed compared with that for site B.
Acknowledgment
Fig. 10. Arrhenius plots of PL decay time of the dominant site for AlGaN:(Eu, Mg).
Table 1 Activation energy of AlGaN:(Eu, Mg) with different Al compositions. Al composition (%) 27.6 Site Activation energy (meV)
0
6.6
6.6
A 161
A 198
B 63
16.6 B 68
B 113
investigated. The PL decay time of site A for XAl values of 0% and 6.6% and those of site B for XAl greater than 6.6% were evaluated. The Arrhenius plots of PL decay time of each site for AlGaN:(Eu, Mg) are shown in Fig. 10. In the phonon-assisted back-transfer model suggested in Ref. [28], the temperature dependence of PL decay time can be expressed as
⎛ Ea ⎞ 1 1 1 exp ⎜ − ⎟ = + ⎝ kT ⎠ τx (T ) τ0 τBT
(2)
where τ0 is the decay time at LT, τBT is a fitting parameter, and Ea is the activation energy, which is the difference in energy between the 5D0 level and the trap level in the host related to the back-
This work was partly supported by a Grant-in-Aid Scientific Research (C) #26420271 from the Ministry of Education, Culture, Sports, Science and Technology in Japan and the Toyoaki Scholarship Foundation.
References [1] P.N. Favennec, H. L'Haridon, M. Salvi, D. Moutonnet, Y.L. Guillou, Electron. Lett. 25 (1989) 718. [2] A. Wakahara, Opt. Mater. 28 (2006) 731. [3] K. Wang, R.W. Martin, K.P. O'Donnell, V. Katchkanov, E. Nogales, K. Lorenz, E. Alves, S. Ruffenach, O. Briot, Appl. Phys. Lett. 87 (2005) 112107. [4] S. Morishima, T. Maruyama, K. Akimoto, J. Cryst. Growth 209 (2000) 378. [5] E.E. Nyein, U. Hömmerich, C. Munasinghe, A.J. Steckl, J.M. Zavada, Mater. Res. Soc. Symp. Proc. 866 (V3) (2005) 5. [6] R. Wang, A.J. Steckl, J. Cryst. Growth 312 (2010) 680. [7] M. Zhong, A.J. Steckl, Appl. Phys. Express 3 (2010) 121003. [8] J.-H. Park, A. Wakahara, H. Okada, H. Sekiguchi, A. Tiwari, Y.-T. Kim, J. Song, J.-H. Lee, J. Jhin, Jpn. J. Appl. Phys. 50 (2011) 031003. [9] M. Pan, A.J. Steckl, Appl. Phys. Lett. 83 (2003) 9. [10] Y.Q. Wang, A.J. Steckl, Appl. Phys. Lett. 82 (2002) 502. [11] H. Okada, K. Takemoto, F. Oikawa, Y. Furukawa, A. Wakahara, S. Sato, T. Ohshima, Phys. Status Solidi C 6 (2009) S631. [12] A. Nishikawa, T. Kawasaki, N. Furukawa, Y. Terai, Y. Fujiwara, Appl. Phys. Express 2 (2009) 071004.
66
M. Kanemoto et al. / Journal of Luminescence 166 (2015) 60–66
[13] H. Sekiguchi, Y. Takagi, T. Otani, R. Matsumura, H. Okada, A. Wakahara, Jpn. J. Appl. Phys. 52 (2013). [14] R. Dahal, C. Ugolini, J.Y. Lin, H.X. Jiang, J.M. Zavada, Appl. Phys. Lett. 97 (2010) 141109. [15] K. Wang, K.P. O'Donnell, B. Hourahine, R.W. Martin, I.M. Watson, K. Lorenz, E. Alves, Phys. Rev. B 80 (2009) 125206. [16] U. Hömmerich, E.E. Nyein, D.S. Lee, J. Heikenfeld, A.J. Steckl, J.M. Zavada, Mater. Sci. Eng. B 105 (2003) 91. [17] Y. Nakanishi, A. Wakahara, H. Okada, A. Yoshida, T. Ohshima, H. Itoh, T. Shibata, M. Tanaka, Phys. Status Solidi C 0 (2003) 2623. [18] E.E. Nyein, U. Hommerich, J. Heikenfeld, D.S. Lee, A.J. Steckl, J.M. Zavada, Appl. Phys. Lett. 82 (2003) 1655. [19] A. Fleischman, C. Munasinghe, A.J. Steckl, A. Wakahara, J. Zavada, V. Dierolf, Appl. Phys. B 97 (2009) 607. [20] L. Bodiou, A. Oussif, A. Braud, J.-L. Doualan, R. Moncorge, K. Lorenz, E. Alves, Opt. Mater. 28 (2006) 780.
[21] I.S. Roqan, K.P. O'Donnell, R.W. Martin, P.R. Edwards, S.F. Song, A. Vantomme, K. Lorenz, E. Alves, M. Bockowski, Phys. Rev. B 81 (2010) 085209. [22] N. Woodward, J. Poplawsky, B. Mitchell, A. Nishikawa, Y. Fujiwara, V. Dierolf, Appl. Phys. Lett. 98 (2011) 011102. [23] K. Takahei, A. Taguchi, J. Appl. Phys. 74 (1993) 1979. [24] Y. Takagi, T. Suwa, H. Sekiguchi, H. Okada, A. Wakahara, Appl. Phys. Lett. 99 (2011) 171905. [25] H. Sekiguchi, Y. Takagi, T. Otani, H. Okada, A. Wakahara, J. Appl. Phys. 113 (2013) 013105. [26] D.-G. Lee, A. Nishikawa, Y. Terai, Y. Fujiwara, Appl. Phys. Lett. 100 (2012) 171904. [27] J.K. Mishra, T. Langer, U. Rossow, S. Shvakov, A. Wieck, A. Hangleiter, Appl. Phys. Lett. 102 (2013) 061115. [28] F. Priolo, G. Dranzo, S. Coffa, A. Carnera, Phys. Rev. B 57 (1998) 4443.