Optical properties of high-temperature annealed Eu-implanted GaN

Optical Materials 28 (2006) 797–801 www.elsevier.com/locate/optmat

Optical properties of high-temperature annealed Eu-implanted GaN K. Wang a,*, R.W. Martin a, E. Nogales a, V. Katchkanov a, K.P. OÕDonnell a, S. Hernandez a, K. Lorenz b, E. Alves b, S. Ruffenach c, O. Briot c a

Department of Physics, University of Strathclyde, Glasgow G4 0NG, United Kingdom b ITN, Estrada Nacional 10, 2686-953 Sacave´m, Portugal c GES, Universite´ de Montpellier II, 34095 Montpellier, France Available online 10 November 2005

Abstract A 10 nm thick epitaxially grown AlN cap has been used to protect the surface of a GaN epilayer both during Eu ion implantation and the subsequent high-temperature annealing. The 15 K photoluminescence (PL) intensity of the intra-4f Eu transition increases by two orders of magnitude when the annealing temperature is increased from 1000 to 1300 C. High-resolution PL spectra reveal that the emission lines due to the 5D0–7F2 transition exhibit different dependencies on the annealing temperature in the studied annealing range. PL excitation measurements demonstrate band edge absorption by the GaN host at 356 nm, together with a broad excitation band centred at 385 nm. The PL spectra of the 5D0–7F2 transition selectively excited by above band-gap absorption and by this broad excitation band are noticeably different. The first peak at 620.8 nm is suppressed when exciting below the GaN band gap. This demonstrates differing energy transfer processes for the different Eu luminescent peaks and is direct evidence for at least two kinds of different Eu sites in the host with distinct optical activation. Temperature dependent PL and PLE demonstrate that one of the two Eu-centres does not contribute to the room temperature luminescence.  2005 Elsevier B.V. All rights reserved.

1. Introduction In the last few years, rare earth-doped (RE) GaN has become scientifically and technologically interesting because its light emission is less sensitive to temperature than RE-doped narrow band gap semiconductor [1–3]. The most prominent Eu-related luminescence occurs in the red spectral range around 621 nm and is ascribed to the intra-4f shell 5D0–7F2 transition. Eu-doped GaN is particularly attractive for realization of red light emitters because although InGaN-based devices are suitable UVblue-green light emitters they are much less efficient for red emission [4]. However, while GaN:Eu devices have been demonstrated [1–3,5], the emitting centres and the detailed energy transfer process from host to dopant in GaN are still open questions.

*

Corresponding author. E-mail address: [email protected] (K. Wang).

0925-3467/$ - see front matter  2005 Elsevier B.V. All rights reserved. doi:10.1016/j.optmat.2005.09.026

Nyein et al. have inferred the existence of different Eurelated optically active centres from the observation that the PL decay dynamics and thermal quenching process depend on the energy of excitation [6]. PL excitation (PLE) measurements on the 622 nm emission by the same group revealed five absorption lines in the spectral region from 570 to 590 nm associated with the 7F0–5D0 transition. Selective excitation of PL by resonant pumping into these lines changed the envelope of the 620 nm spectral emission, although individual lines were not resolved [7]. PL excitation is also a useful technique to study the excitation transfer from the host GaN, via such processes as carriermediated excitation, excitation by excitons, and defectmediated excitation. For example, PLE studies of GaN:Eu at 300 K identified the existence of a broad sub-gap absorption band at 400 nm [6]. Lee and Peng et al. reported temperature dependent and excitation wavelength dependent PL decay transients and they are attributed to different energy transfer pathways [8,9]. By using the systematic approach of combined excitation-emission spectroscopy (CEES), Dierolf et al. present a detailed assignment of

K. Wang et al. / Optical Materials 28 (2006) 797–801

PL and PLE peaks for two dominant Er3+ sites and several minor ones in GaN:Er [10]. The identified sites differ strongly in their excitation and energy transfer efficiencies. Here, we report the effect of annealing, PL excitation and selectively excited PL results on the fine structure of the 5D0–7F2 multiplet in Eu:GaN. Temperature dependent PL and PLE are also investigated. The results show that the four main lines around 621 nm resolve into two distinct pairs. One pair is selectively excited via a broad absorption band centred near 385 nm indicating that different excitation mechanisms and different types of optically active Eu ions co-exist within the sample.

620.8 nm 5

D0

7

F2

621.7 nm

622.5 nm

PL intensity (a.u.)

798

5

D0

7

F1

o

1300 C

600.2

600.8

× 5

1100 oC

× 50 1000 oC

2. Experimental 590

Two lm thick GaN epilayers were grown by MOVCD on c-plane sapphire at 1090 C following a low temperature GaN buffer layer. They were capped with a 10 nm thick monocrystalline layer of AlN, grown epitaxially at 1160 C. Eu ions were implanted at normal incidence through the cap at 500 C using an energy of 300 keV and fluences of 1 and 2 · 1015 at/cm2. The AlN cap allowed the implanted samples to be annealed for 20 min at temperature from 1000 to 1300 C in a conventional tube furnace with a moderate overpressure of 4 bar nitrogen gas. The structural properties of the samples and the increased optical activation of the Eu ions compared to samples annealed at low temperature were reported previously [11]. The samples were mounted in a closed-cycle helium cryostat and cooled from room temperature to 15 K. The 325 nm line of a 5 mW He–Cd laser was used to study the effect of annealing on the emission spectra, and the PL was dispersed by a 0.125 m spectrograph and detected by a CCD camera. The excitation source for PLE was a 1000 W short arc xenon lamp combined with a 0.25 m focal length monochromator. The incident power on the sample averages 0.6 mW on a surface area of about 2 · 5 mm2. The luminescence was analysed by a 0.67 m focal length monochromator and detected using a cooled photomultiplier tube (PMT). PLE spectra were obtained by fixing the detection monochromator at the peak of a particular emission line and scanning the excitation monochromator. The spectral resolution in the emission spectra was 0.1 nm and that in the excitation spectra was 4 nm. Selectively excited PL spectra were obtained using the Xe lamp with the excitation monochromator set at particular wavelengths. All the PLE spectra were corrected for the output of the xenon lamp and the throughput of the excitation monochromator, and normalized to the intensity of a PLE peak at 356 nm, which was a common feature of all PLE spectra. 3. Results and discussion 3.1. Effect of annealing on luminescence of Eu:GaN A series of samples with the same fluence of 1 · 1015 at/ cm2 have been annealed at 1000, 1100 C, and 1300 C,

595

600

605

610

615

620

625

630

Wavelength (nm) Fig. 1. 15 K PL spectra of AlN-capped GaN samples implanted with 1 · 1015 Eu/cm2 and annealed at 1000, 1100, 1300 C. The 325 nm line of a He–Cd laser was used as excitation source.

respectively. Fig. 1 shows the PL spectra around the main red emission of samples annealed at different temperatures. The emission intensity is increased about 2 order of magnitude when the anneal temperature increased from 1000 to 1300 C. Notably, the peak at 620.8 nm cannot be seen after annealing at 1000 C but became the strongest one after annealing at 1300 C. On the other hand, the middle peak at 621.7 nm is the strongest in the spectra after annealing at 1000 and 1100 C. Moreover, for the emission peaks due to 5D0–7F1, the peak at 600.2 nm increases much faster than the peak at 600.8 nm as the anneal temperature increases and dominates the latter one after annealing at 1300 C. Thus, the PL intensity of these peaks have different dependencies on the anneal temperature. The results suggest that these emission peaks are from different Eu-related centres. Annealing increases the emission efficiency of these different centres in different ways. The Eu centre responsible for the first main peak at 620.8 nm needs more thermal energy to become optically active than the other peaks and luminescence from this Eu centre requires higher temperature annealing (1300 C). 3.2. Selectively excited PL and PL excitation spectra A typical Eu3+ PL spectrum for excitation at 356 nm (3.48 eV) is shown in Fig. 2(a). The PL spectrum shows three main peaks, at 620.8, 621.7, 622.5 nm and three minor peaks are resolved at 617.2 nm, 618.7 nm and 619.3 nm on the shorter wavelength side of the main peaks. Shoulders also appear in the wavelength range 623–625 nm. The first main peak at 620.8 nm is the dominant emission with FWHM 0.4 nm, and the third line at 622.5 nm is weak. More interestingly, the middle line is seen to consist of two components with a peak at 621.7 nm and a shoulder at 621.9 nm. In previously published works, only the three

K. Wang et al. / Optical Materials 28 (2006) 797–801

620.8 nm 5

D0

7

621.7 nm

F2

th

4 line

PL intensity (a.u.)

(a) Excited by 356 nm (b) Excited by 385 nm 622.5 nm

(a) (b)

× 10

616 617 618 619 620 621 622 623 624 625

Wavelength (nm) Fig. 2. 15 K PL spectra (around 621 nm) of AlN-capped GaN implanted with 2 · 1015 Eu/cm2, after annealing at 1300 C. Excited under (a) above band gap and (b) below band gap conditions by a xenon lamp.

main peaks with a broadened middle peak have been reported and the splitting of the central peak has not been clear [7,8]. Our results show the emission due to the 5 D0–7F2 transition to consist of 4 main emission lines [12]. Normalized PLE spectra taken at the three main emission peaks (Fig. 3) clearly demonstrate above-gap excitation (<355 nm) and an exciton-like absorption peak at 356 nm (3.48 eV). The main difference between the three PLE traces is the appearance of a broad excitation band from 360 to 410 nm centred at about 385 nm for the emission lines at 621.7 nm (unresolved doublet) and 622.5 nm (singlet). Some weak structures can also be observed within this band at 372 and 385 nm. The PLE spectrum detected at 621.7 nm collects light from two emission peaks to produce an excitation spectrum, which is roughly the average of the other two. Hence, only the line pair at 621.7 and 622.5 nm is excited in the broad band. It is interesting to note further that PL spectra of luminescent

Eu:GaN samples generally show a set of near-band-edge luminescence peaks in the wavelength range from 370 to 400 nm [13]. Fig. 2(b) confirms that the PL spectrum excited below the band gap is markedly different from the spectrum excited at 356 nm. The line at 620.8 nm is now the weakest of the four and the peak at 622.5 nm is enhanced. Moreover, the middle peak becomes much sharper and is slightly blue shifted to 621.6 nm. This is due to the selective enhancement of one of the pair of unresolved lines. The minor peaks in the shorter wavelength range from 617 to 620 nm disappear completely from the PL spectrum, and the intensity of the shoulders in the longer wavelength range from 623 to 625 nm is also reduced. We have also investigated selectively excited PL in the regions near 602 nm (5D0–7F1) and 664 nm (5D0–7F3) and seen effects similar to those for 5D0–7F2, with the relative strength of the lines within the observed multiplets strongly modified when the excitation energy is changed from above to below the GaN band gap. Fig. 4(a) shows that the two peaks at 600.3 and 601.6 nm dominate in the spectrum when excited at 356 nm. On the other hand, they are not visible when excited below band gap at 385 nm as shown in Fig. 4(b). Instead, two new peaks at 601.0 and 602.2 nm appear. Normalized PLE spectra (Fig. 5) detected at these 4 peaks demonstrate above band gap excitation and below band gap excitation. The broad excitation band from 360 to 410 nm are clearly shown when detected at 601.0 and 602.2 nm emission peaks. The PLE and PL results discussed above provide information about the different energy transfer processes in GaN:Eu. Firstly, band to band GaN absorption followed by carrier-mediated energy transfer to the Eu ions leads to the intra-4f shell transition with subsequent light emission. Secondly, excitation of luminescence from the Eu ions via a

5

1

D0

At 620.8 nm At 621.7 nm At 622.5 nm

(a) Excited by 356 nm (b) Excited by 385 nm

7

F1

PL intensity

Normalized PLE intensity (a.u.)

799

(a) 0

(b)

330 340 350 360 370 380 390 400 410 420 430 440

Wavelength (nm) Fig. 3. 15 K normalized PLE spectra of AlN-capped GaN implanted with 2 · 1015 Eu/cm2, after annealing at 1300 C. A xenon lamp was used to scan through the wavelength range and the detector was fixed at the three main peaks around 621 nm.

598

599

600

601

602

603

604

Wavelength (nm) Fig. 4. 15 K PL spectra (around 601 nm) of AlN-capped GaN implanted with 2 · 1015 Eu/cm2, after annealing at 1300 C. Excited under (a) above band gap and (b) below band gap conditions by a xenon lamp.

800

K. Wang et al. / Optical Materials 28 (2006) 797–801

At 600.3 nm At 601.0 nm At 601.6 nm At 602.2 nm

Normalized PLE intensity (a.u.)

1

PL intensity (a.u.)

15K 50K 75K 100K

150K

0 340

360

380

400

420

440

Wavelength (nm)

200K

Fig. 5. 15 K normalized PLE spectra of AlN-capped GaN implanted with 2 · 1015 Eu/cm2, after annealing at 1300 C. A xenon lamp was used to scan through the wavelength range and the detector was fixed at the 4 peaks around 601 nm.

250K 297K 616

618

620

622

624

Wavelength (nm)

process involving the GaN exciton peak at 356 nm is a feature common to all the PLE spectra, as shown by Fig. 2. Thirdly, a pair of PL emission lines at 621.6 and 622.5 nm can be excited below band gap at 385 nm, indicating an excitation pathway involving defect levels or complexes associated with a fraction of Eu ions. It is interesting to note that a broad absorption band at 400 nm, observed for Erimplanted GaN [14], can be enhanced by Mg co-doping which selectively increases the intensity of emission at 1554.8 nm from one of the Er3+centres in GaN [15]. For Eu-doped GaN, the control of defect levels within the GaN band gap by, for example, co-doping could possibly be used to increase the efficiency of Eu-related luminescence.

Fig. 6. PL spectra excited by 385 nm light from 15 to 297 K. The slits were reduced at lower temperature than 175 K due to strong signal but the spectral pattern will not change because of different band pass. (This is the same for Fig. 7.)

Fig. 7 shows temperature dependent PLE spectra detected at 622.5 nm peak. The exciton-like peaks in the spectra exhibit a typical GaN PL exciton quenching process and a red shift from 356 to 364 nm. The excitation

15K 50K

75K

PL and PLE spectroscopy has been carried out from 15 to 297 K. Temperature dependent measurements demonstrate different thermal quenching rates for the three main peaks excited by He–Cd laser at 325 m in our previous work [13]. The PL peak at 620.8 nm quenches much faster than the other two peaks when the sample is excited above band gap. Below band gap excitation (at 385 nm) separates the emission pair at 621.7 and 622.5 nm from the mixed PL spectra. Fig. 6 shows the quenching processes of this pair when excited at 385 nm by Xe lamp. The emission intensity at 297 K is 20% of that at 15 K. The PL peak at 620.8 nm does not contribute to the room temperature luminescence. The temperature dependence of the two peaks is slightly different. The 621.7 nm peak is stronger than the 622.5 nm peak at low temperature, but becomes weaker when the temperature is increased close to room temperature.

100K

PLE intensity (a.u.)

3.3. Temperature dependent PL and PLE measurements

125K

150K

200K 225K

×3

250K 297K

340

360

380

400

420

440

Wavelength (nm) Fig. 7. PLE spectra detected at 622.5 nm from 15 to 297 K. The spectral intensities are multiplied by a factor of 3 for higher temperatures than 150 K.

K. Wang et al. / Optical Materials 28 (2006) 797–801

band below band gap can be observed in all temperature ranges and experiences a slower decrease with increasing temperature than the excitonic peak. The PL intensity excited below band gap via Eu-related defects is only 35% of that excited at 356 nm at 15 K. Nevertheless, this ratio increases to 65% at room temperature. The results suggest that excitation of Eu intra-4f shell transitions through Eu-related defects is one of the key issues to improve the red emission efficiency. 4. Conclusion Photoluminescence (PL) and PL excitation spectra have been studied for Eu-implanted and high-temperature annealed GaN with an AlN capping layer. The PL intensities of the main red emission peaks have different dependencies on the anneal temperature. The dependence of the PL spectra on the excitation mode (above band gap excitation and below band gap excitation) reveals contributions of at least two kinds of optically active Eu3+ centres to the emission near 622 nm. PL excitation spectra demonstrate that one kind of Eu centre is excited following a broad absorption band centred at around 385 nm. Temperature dependent PL measurements demonstrate different quenching behaviour for distinct Eu3+ luminescent centres. Temperature dependent PLE measurements indicate that Eu-related defect excitation band below band gap plays an important role in red emission at room temperature.

801

References [1] A.J. Steckl, J.M. Zavada, MRS Bull. 24 (1999) 33. [2] A.J. Steckl, J. Heikenfeld, D.S. Lee, M.J. Garter, C.C. Baker, Y.Q. Wang, R. Jones, IEEE J. Sel. Top. Quant. Electron. 8 (2002) 749. [3] J. Heikenfeld, M. Garter, D.S. Lee, R. Birkhahn, A.J. Steckl, Appl. Phys. Lett. 75 (1999) 1189. [4] T. Mukai, H. Harimatsu, S. Nakamura, Jpn. J. Appl. Phys. 37 (1998) L479. [5] S. Morishima, T. Maruyama, M. Tanaka, Y. Masumoto, K. Akimoto, Phys. Status Solidi (a) 176 (1999) 113. [6] E.E. Nyein, U. Ho¨mmerich, J. Heikenfeld, D.S. Lee, A.J. Steckl, J.M. Zavada, Appl. Phys. Lett. 82 (2003) 1655. [7] U. Ho¨mmerich, E.E. Nyein, D.S. Lee, J. Heikenfeld, A.J. Steckl, J.M. Zavada, Mater. Sci. Eng. B 105 (2003) 91. [8] Chang-Won Lee, H.O. Everitt, D.S. Lee, A.J. Steckl, J.M. Zavada, J. Appl. Phys. 95 (2004) 7717. [9] H.Y. Peng, Chang-Won Lee, H.O. Everitt, D.S. Lee, A.J. Steckl, J.M. Zavada, Appl. Phys. Lett. 86 (2005) 051110. [10] V. Dierolf, C. Sandmann, J. Zavada, P. Chow, B. Hertog, Appl. Phys. Lett. 95 (2004) 5464. [11] K. Lorenz, U. Wahl, E. Alves, S. Dalmasso, R.W. Martin, K.P. OÕDonnell, S. Ruffenach, O. Briot, Appl. Phys. Lett. 85 (2004) 2712. [12] K. Wang, R.W. Martin, K.P. OÕDonnell, V. Katchkanov, E. Nogales, K. Lorenz, E. Alves, S. Ruffenach, O. Briot, Appl. Phys. Lett. 87 (2005) 112107. [13] K.P. OÕDonnell, V. Katchkanov, K. Wang, R.W. Martin, P.R. Edwards, B. Hourahine, E. Nogales, J.F.W. Mosselmans, B. De Vries, Mater. Res. Symp. Proc. 831 (2004) E9.6.1. [14] S. Kim, S.J. Rhee, X. Li, J.J. Coleman, S.G. Bishop, P.B. Klein, J. Electron. Mater. 27 (1998) 246. [15] S. Kim, S.J. Rhee, X. Li, J.J. Coleman, S.G. Bishop, Appl. Phys. Lett. 76 (2000) 2403.