Open Access
Add to BibSonomyAbstract
Eu ions in situ doped in GaN with V/III ratios varying from 3200 to 9600 have been investigated using resonant site-selective photoluminescence (PL), power dependent cathodoluminescence (CL), and a unique electron beam power dependent dual excitation experiment combining the techniques of PL and CL. The results of these experiments reveal the role of defects in the electronic excitation of Eu ions and the link between the GaN host and Eu ion dopants. The relative number of beneficial defects present in each sample for a majority Eu site (Eu1) and a specific secondary site (Eu2) are revealed. Also, a room temperature activated non-radiative recombination pathway linked to a specific, sample dependent Eu2 excitation pathway is identified. Unlike conventional GaN LEDs, Eu:GaN device performance does not rely completely on crystalline quality, but on the presence of specific excitation enhancing defects and the absence of non-radiative de-excitation channels.
© 2013 Optical Society of America
1. Introduction
Rare earth doping of gallium nitride (GaN) is a promising approach for the realization of light emitting diodes (LEDs) and electrically pumped lasers in the wavelength ranges characterized by the RE emission (i.e. 1.5 μm for Er3+ 1064 nm for Nd3+, and 620 nm for Eu3+) [1–8]. The red emission of Eu is of particular interest since the efficiencies of red quantum well GaN LEDs are very low. High efficiency red emission from a GaN based material will allow for a single crystal white LED that can be used for solid state lighting applications and full LED displays. However, devices that incorporate Eu in the active layer show a saturation effect at injection levels above 20 mA and the light output efficiency decreases substantially with higher injection currents. Recently, major improvements in the maximum output power and efficiency has been achieved by in situ Eu doping using an organometallic vapor phase epitaxy (OMVPE) growth method. This method has allowed for external quantum efficiencies of ~0.6% and a record light output power of ~50 μW [2]. These advances have been possible with the optimization of growth parameters such as pressure, temperature, and thickness of the active region [2–5]. Through these variations, the defects within the crystal structure are modified. These defects can be beneficial by directing electron and hole recombination energy to Eu ions, while others can be detrimental by facilitating unwanted relaxation channels before and after the Eu ion has been excited. Understanding and controlling these defect-related processes is essential to achieve commercially practical output powers of at least 1 mW.
2. Experimentation
The ~300 nm thick active layers of the samples used in this study were grown by OMVPE on sapphire substrates with a GaN buffer layer. The V/III fluxes during growth for each sample were varied to produce Eu:GaN active layers with 3200, 6400, and 9600 V/III ratios. To avoid confusion, these samples have been labeled Gan320, GaN640, and Gan960. They have Eu concentrations of 6 x 1019, 4 x 1019, and 2 x 1019 cm−3 respectively.
In our analysis of the defect related processes of each sample, there has been four different types of defects distinguished according to their functions: (1) defects that directly perturb the Eu ion excitation and emission energies leading to spectrally distinguishable Eu sites, (2) defects that assist in energy transfer from electron-hole pairs (EHPs) to Eu ions referred to as excitation enhancing defects (EEDs), (3) defects that capture energy from EHP recombination, but do not transfer any energy to the Eu ion, and (4) defects that give rise to non-radiative decay of the Eu ion after it has been excited. Enhancing the number of defect type 2 will further improve the device output powers after the injection of electron and holes into the active layer.
In order to clarify the role and relative numbers of particular defects in the presence of other defects, they have been correlated with the number of specific Eu sites and overall emission efficiencies. This was possible by combining the results from specially designed emission experiments including photoluminescence (PL), cathodoluminescence (CL), and combined electron/laser beam (CELB) excitation experiments.
To do these three measurements in an efficient and reliable fashion, a scanning confocal luminescent microscope (SCLM) was developed inside a scanning electron microscope (SEM) that enables their simultaneous execution. Figure 1 shows the SCLM as it sits on an Oxford low-temperature stage that fits inside a JEOL 6400 equipped with a JEOL SM-40010 beam blanking unit for time resolved CL (TRCL) experiments. The laser light is guided to the instrument through fibers and directed to the sample by a beam splitter, mirrors, and lenses. The collected emission is directed into a second fiber by similar optics and guided into a Princeton Instruments Acton Spectra Pro 2300i spectrometer equipped with a Princeton Instruments Spec-10 liquid nitrogen cooled CCD array. The properties of a confocal microscope define a small collection area (r = 2 µm for this setup) that is overlapped with the laser excitation and a decreased depth of focus. In this case, the depth of focus is limited by the 300 nm active layer thickness. The result is a good electron and laser beam overlap in the effective collection area suitable for our combined excitation experiments [9].
Fig. 1 (left) A schematic of the CELB apparatus built for this experiment. (right) The CELB apparatus as it sits on an Oxford low temperature stage to be inserted into a JEOL6400 SEM. The fiber closest to the camera (1) is a 3.5 µm fiber that guides laser light to the microscope. This fiber is attached to a Thorlabs fiberport, which collimates and directs the beam into a beam splitter (2). The beam splitter reflects the light in the direction of two mirrors (3) that guide the collimated beam through an objective lens (4), which then focuses the beam onto a sample. The light emitted by the sample, resulting from either laser or electron beam excitation, is collected and collimated by the objective lens (4). The collimated light is directed through the beam splitter by the two mirrors (3) and coupled into a 6 µm collection fiber (5) by a Thorlabs fiberport coupler.
For the PL experiments, combined excitation emission spectroscopy (CEES) techniques were used to reveal the relative concentrations of different Eu incorporations sites. CEES utilizes a tunable laser source that is stepped by a specific wavelength increment (~0.02 nm for this study), while the emission spectra are recorded for each excitation step [10]. These measurements were done at liquid helium temperatures (~4K). Power dependent and time resolved CL (TRCL) experiments were used to identify non-radiative de-excitation channels at room temperature. The CELB experiment was used to determine the relative number of excitation enhancing defects (EEDs) (the 2nd type of defect listed previously) with a spatially overlapped below band gap laser (2.16 eV) and 10 kV e-beam.
The e-beam was defocused to a spot size of ~35 µm to ensure that the entire collectible active layer defined by the confocal microscope was excited. The laser photon energy was chosen to interact with Eu-coupled defect states that directly excite the Eu ions. It has been shown that such laser excitations decrease the CL emission intensity due to the ionization of these defect states before Eu excitation [11]. The magnitude of the decrease in signal is directly related to the number of EEDs being ionized by the laser excitation.
Therefore, the relative number of EEDs can be experimentally measured by adjusting the electron beam power density while simultaneously exciting the sample with a constant laser power.
3. Results
3.1 Room Temperature CL
The room temperature power dependent CL results for GaN320, GaN640, and GaN960 are shown in Fig. 2. As expected, GaN960 shows the lowest emission intensity due to its low Eu concentration. However, GaN640 outputs 3/2 times more light than GaN320 under saturated CL power densities despite having 2/3 the Eu concentration. These results suggest that the defect configuration of GaN640 is more beneficial for Eu excitations than that of GaN320. A series of CEES experiments were performed to further elucidate the superior performance of the GaN640 sample.
Fig. 2 The total integrated CL intensity as a function of e-beam current for a 1.80 - 2.25 eV emission at room temperature. A 10 kV accelerating voltage and a 35 µm spot size was used to excite the entire volume of the Eu:GaN layer defined by the collection of the confocal microscope.
3.2 Low Temperature PL
The zero-phonon CEES data with laser excitations stepped between 2.101-2.117 eV and emissions collected between 1.974-2.013 eV shown in Fig. 3 correlate to the 7F0 to 5D0 excitations and 5D0 to 7F2 emissions of the Eu incorporation sites. Eight different sites are labeled for GaN320 consistent with Woodward et al. [10]. It is clear that perturbations are causing more pronounced energy splittings of the 7F2 state, a decrease in energy from the 7F0 to 5D0 excitation, as well as extra sites that do not exist in conventionally grown Eu:GaN. Also, fluorescence line narrowing effects (apparent by the diagonal strips) become more prominent with higher V/III ratios, which suggests that higher V/III ratios result in more crystalline disorder. Combining the CL results with the zero-phonon CEES results indicate that, for these samples, the device performances do not completely correlate with the crystalline quality and the absence of defects.
Two of the eight sites, labeled OMVPE 4 and 7, contribute to a majority (> 90%) of the total electronic red light emission, and have been relabeled to Eu1 and Eu2 to be consistent with literature [12]. Eu2 can be excited throughout the entire CEES excitation range for GaN320, while it can only be excited with a phonon-assisted resonant excitation for GaN640 and GaN960 (not shown in Fig. 3). Therefore, Eu2 may or may not be coupled to a deep level gap state depending on the defect configuration of the particular sample. The CELB and TRCL experiments will show that the coupling of Eu2 to this deep level state is detrimental to the device performance.
The samples’ Eu concentration, total integrated room temperature CL emission, and the Eu1 and Eu2 phonon-assisted resonant PL intensities are compared in Fig. 4. The phonon-assisted PL excitation regime was used for this comparison because it gives the most reliable data for the relative numbers of Eu centers [10]. The strong correlation with the Eu concentration and the phonon-assisted resonant Eu1 emission intensities is indicative of a Eu1 majority center in Eu:GaN [13]. Also, the strong correlation of the phonon-assisted resonant Eu2 emission with the total integrated CL intensity suggests that the Eu2 concentration is essential to the device performance.
Fig. 4 The comparison of the Eu concentrations to the Eu1 phonon-assisted resonant PL intensities and the room temperature total integrated CL intensities to the Eu2 phonon-assisted resonant PL intensities for the samples in this study.
3.3 Combined excitation experiment
An e-beam power dependent CELB experiment with a constant laser power was performed to investigate the number of EEDs associated with Eu1 and Eu2 along with their excitation efficiencies at 125 K. Two spectra were recorded at each e-beam current, one with and one without laser excitation. The emissions of Eu1 and Eu2 with and without laser excitation were subtracted at each e-beam current interval to give the decrease in CL emission resulting from laser induced defect ionization before Eu excitation. The results are depicted in Fig. 5.
Fig. 5 The decrease in CL emission with additional below band gap laser excitation (2.16 eV) as a function of the e-beam current. The solid lines are curves fit to the CL emission subtracted from the CL emission with additional laser excitation data using Eq. (1).
The slopes of the curves in Fig. 5 are steep and positive for lower e-beam currents, while they become less steep and negative for higher currents. At lower e-beam currents, the EHP generation rate is low, which decreases the probability for Eu excitation by EEDs. Increasing the e-beam current increases the EHP generation rate and EED excitation rate. Therefore, a higher concentration of excited EEDs is available for photon ionization at any instant in time. This leads to a more drastic reduction in CL emission with additional laser excitation for higher e-beam currents, which explains the positive slope in Fig. 5. In this scenario, the photon induced EED ionization rate must be much greater than the rate at which EEDs can transfer energy to the Eu sites. For higher e-beam currents, the slope decreases and becomes negative. The decreasing slope is a result of a saturation of excited EEDs by EHPs. As the EEDs reach saturation, the EHP-EED excitation rate overcomes the laser ionization rate, and the EEDs are re-excited instantaneously after laser ionization. This effect can be seen more prevalently for the Eu2 peak since the Eu2 excitation saturates much before Eu1. The negative slope is only shown for the Eu2 peak in the GaN320 sample. The same results are expected with higher currents for the other samples. A lower current range was used in this study to inhibit unwanted e-beam induced artifacts, which limited the experiment.
In order to obtain more quantitative results, the processes of EED excitation, ionization, energy transfer, and Eu emission decay can be translated into a rate equation model, which yields the following equation when solved:
is the magnitude of the decrease in Eu CL emission intensity with additional laser excitation, is the number of Eu coupled EEDs, is the rate Eu ions are excited by EHPs, I is the e-beam current and is proportional to the EHP generation rate, is the Eu decay rate, is the laser induced trap ionization rate, and is a constant that describes all instrumentation factors that determine the collection efficiency, the EHP generation rate with respect to I, and the fact that the decay of Eu is not purely radiative. is known from time resolved CL (TRCL) measurements, is proportional to the laser intensity and set constant, and A was set constant because the measurements were done at low temperature. Setting these constants and fitting the CELB data to Eq. (1) reveals the relative numbers of EEDs () and excitation rates () centered at the Eu1 (1.991 eV) and Eu2 (1.996 eV) emission peaks for each sample. The resulting fit curves are displayed in Fig. 5. The exact numbers for and are unable to be determined by these methods, but the relative numbers can be compared between each sample. To this end, each parameter was divided by the lowest value to give a direct correlation between the parameters. Table 1 shows the obtained parameters used for each sample and center. The reliability of the fits and experimental procedure is reflected in the fact that the sample independent parameters (, , and ) were kept constant.
Table 1. Relative Values for NEED and kexc
There are significant correlations between the values of for Eu1 and Eu2, the room temperature saturated CL total integrated intensities, and the Eu2 phonon-assisted resonant PL intensities for each sample, which is depicted in Fig. 6. The strongest correlations are realized between the Eu2 resonant PL intensity and for Eu2 as well as the total integrated intensity and for Eu1. The Eu2 phonon-assisted resonant PL intensities andnumbers of EEDs correlation suggest that it is more energetically favorable for EEDs to be spatially close to the Eu2 center. This leads to the high probability for electronic excitation measured for Eu2 centers. The correlation of EEDs for Eu1 and that of Eu2 suggests that the excitation of Eu1 centers by EHPs heavily depends on the same type of EED. However, it may be less likely for a Eu1 center to be spatially close to an EED. A larger spatial separation would lead to lower excitation probabilities and, therefore, would allow for the number of Eu2 sites to be a good indicator of the concentration of EEDs essential for Eu1 excitation.
Fig. 6 The comparison of the relative number of Eu1 EEDs to the room temperature total integrated CL intensities and the relative number of Eu2 EEDs to the Eu2 resonant phonon-assisted PL intensities for the three samples in this study. The relative numbers of EEDs were calculated by applying Eq. (1) to the data shown in Fig. 5.
3.4 Time-resolved CL
TRCL measurements have been performed to correlate the room-temperature device performance to non-radiative Eu decay. To do this, the e-beam was turned on and off for 1 ms intervals, while a photomultiplier tube connected to an oscilloscope collected the emitted light. The results are depicted in Fig. 7 for the experiments done at room temperature and 30K. The curves of each sample were artificially moved vertically to compare the slopes, which correlate to the Eu decay time constants. All of the curves exhibit an exponential fit except for the room temperature measurement of GaN320, which shows a double exponential and a much faster time constant. Therefore, GaN320 contains a room temperature activated non-radiative decay center (NRDC) that does not exist in the other samples. The deep-level coupling to Eu2 (Fig. 3) and the fast Eu excitation rate at low temperature ( in Table 1) are characteristics of the same defect responsible for the non-radiative decay at room temperature. This defect improves the rate Eu2 and Eu1 capture excitation from EHPs at low temperature, but is detrimental to the device performance at operating temperatures.
Fig. 7 The TRCL data collected from GaN320, GaN640, and GaN960 at room and low temperature. The Eu emission decay is much faster for GaN320 at room temperature than the other samples and exhibits a double exponential decay.
4. Conclusions
Two types of defects that play a role in the excitation of Eu ions in GaN have been experimentally identified. Both EEDs and room temperature NRDCs enhance the excitation at low temperatures, but room temperature NRDCs lead to non-radiative decay at operational temperatures. Room temperature NRDCs are deep level defects that couple to the Eu2 site and increase low temperature excitation rates (), while EEDs () are not deep defect levels and are not associated with fast low temperature excitation rates. Overall, room temperature NRDCs are unwanted defects, while EEDs are necessary for an increased total Eu emission and improved device performance. The increase of EEDs and decrease in room temperature NRDCs with an increase in the V/III ratio from 3200 to 6400 suggests that these centers are related to a decrease in Ga concentration. Therefore, EEDs may be related to Ga vacancies, which introduce shallow defect levels, and room temperature NRDCs may be related to Ga anti-sites, which introduce deep levels with energies similar to the CEES excitations used in this study [14].
As the V/III ratio exceeds 6400, the total Eu emission suffers as seen for GaN960. The high V/III ratio present in GaN960 decreases the Eu concentration and significantly increases the fluorescence line narrowing indicating a higher crystalline disorder. Therefore, above a threshold limit of V/III ratios, higher concentrations of unwanted/unidentified defects and lower Eu concentrations are detrimental to the overall device performance.
Other growth methods that focus on the passivation of defects that cause fast excitation of Eu ions at low temperature (room temperature NRDCs) and activation of defects that increase the total number of excitable Eu ions (EEDs) need to be tested. A good step would be to research growth parameters that optimize the defect distribution, as seen for GaN640, and have a larger Eu dopant concentration, as seen for GaN320. The success of this material will allow for a high light output Eu:GaN LED that can be used to broaden the GaN LED color spectrum.
Acknowledgments
This work was partly supported by a Grant-in-Aid for Creative Scientific Research (Grant No. 19GS1209) and a Grant-in-Aid for Scientific Research (S) (Grant No. 24226009) from the Japan Society for the Promotion of Science, and partly by the Global Centre of Excellence Program “Advanced Structural and Functional Materials Design” from the Ministry of Education, Culture, Sports, Science and Technology of Japan. All experimental work was preformed at Lehigh University supported by NSF grant ECCS-1140038.
References and links
1. R. Dahal, C. Ugolini, J. Y. Lin, H. X. Jiang, and J. M. Zavada, “1.54 μm emitters based on erbium doped InGaN pin junctions,” Appl. Phys. Lett. 97(14), 141109 (2010). [CrossRef]
2. A. Nishikawa, N. Furukawa, D. Lee, K. Kawabata, T. Matsuno, Y. Terai, and Y. Fujiwara, “Electroluminescence properties of Eu-doped GaN-based light-emitting diodes grown by organometallic vapor phase epitaxy,” MRS Proc. 1342, 9–14 (2011).
3. A. Nishikawa, N. Furukawa, T. Kawasaki, Y. Terai, and Y. Fujiwara, “Improved luminescence properties of Eu-doped GaN light-emitting diodes grown by atmospheric-pressure organometallic vapor phase epitaxy,” Appl. Phys. Lett. 97(5), 051113 (2010). [CrossRef]
4. A. Nishikawa, T. Kawasaki, N. Furukawa, Y. Terai, and Y. Fujiwara, “Electroluminescence properties of Eu-doped GaN-based red light-emitting diode by OMVPE,” Phys. Status Solidi 207(6), 1397–1399 (2010). [CrossRef]
5. A. Nishikawa, T. Kawasaki, N. Furukawa, Y. Terai, and Y. Fujiwara, “Room-temperature red emission from a p-type/europium-doped/n-type gallium nitride light-emitting diode under current injection,” Appl. Phys. Express 2, 071004 (2009). [CrossRef]
6. R. Dahal, C. Ugolini, J. Y. Lin, H. X. Jiang, and J. M. Zavada, “Current-injected 1.54 μm light emitting diodes based on erbium-doped GaN,” Appl. Phys. Lett. 93(3), 033502 (2008). [CrossRef]
7. V. Dierolf, Z. Fleischman, C. Sandmann, A. Wakahara, T. Fujiwara, C. Munasinghe, and A. Steckl, “Combined excitation emission spectroscopy of europium ions in GaN and AlGaN films,” in MRS Proc. 866 (2005), V3.6.
8. D.-G. Lee, A. Nishikawa, Y. Terai, and Y. Fujiwara, “Eu luminescence center created by Mg codoping in Eu-doped GaN,” Appl. Phys. Lett. 100(17), 171904 (2012). [CrossRef]
9. J. Poplawsky and V. Dierolf, “High-resolution confocal microscopy with simultaneous electron and laser beam irradiation,” Microsc. Microanal. 18(6), 1263–1269 (2012). [CrossRef] [PubMed]
10. N. Woodward, A. Nishikawa, Y. Fujiwara, and V. Dierolf, “Site and sample dependent electron–phonon coupling of Eu ions in epitaxial-grown GaN layers,” Opt. Mater. 33(7), 1050–1054 (2011). [CrossRef]
11. J. Poplawsky, N. Woodward, A. Nishikawa, Y. Fujiwara, and V. Dierolf, “Nature and excitation mechanism of the emission-dominating minority Eu-center in GaN grown by organometallic vapor-phase epitaxy,” MRS Proc. 1342, 21–26 (2011).
12. I. Roqan, K. O'Donnell, R. Martin, P. Edwards, S. Song, A. Vantomme, K. Lorenz, E. Alves, and M. Boćkowski, “Identification of the prime optical center in GaN: Eu3+,” Phys. Rev. B 81(8), 085209 (2010). [CrossRef]
13. N. Woodward, J. Poplawsky, B. Mitchell, A. Nishikawa, Y. Fujiwara, and V. Dierolf, “Excitation of Eu3+ in gallium nitride epitaxial layers: Majority versus trap defect center,” Appl. Phys. Lett. 98(1), 011102 (2011). [CrossRef]
14. P. Boguslawski, E. L. Briggs, and J. Bernholc, “Native defects in gallium nitride,” Phys. Rev. B Condens. Matter 51(23), 17255–17258 (1995). [CrossRef] [PubMed]
