Red Electroluminescence from Light Emitting Diodes Based on Eu-Doped ZnO Embedded in p-GaN/Al₂O₃/n-ZnO Heterostructures

The schematic of the whole structure of p-GaN/Al₂O₃/ZnO:Eu/n-ZnO LEDs is depicted in Fig. 1a. Samples were grown at the following schemes on 2-inch c-plane sapphire (0001) substrates. First, a GaN template was grown by metalorganic vapor phase epitaxy (Taiyo Nippon Sanso SR-2000) utilizing Trimethylgallium (TMGa, purity: >99.99995%), ammonia (NH₃, purity: >99.99997%), and bis-cyclopentadienyl magnesium (Cp₂Mg, purity: >99.9999%) as Ga, N and Mg precursors. The growth process was initiated with the growth of low temperature (LT) GaN buffer, ∼1.0 μm-thick undoped-GaN (ud-GaN) and 140 nm-thick p-GaN on sapphire (0001) substrates. Next, after covering half of the GaN template with sapphire substrates, ∼10 nm-thick Al₂O₃ was deposited at the substrate temperature of 300 °C by using atomic layer deposition (SP560-AL, SPLEAD Inc.). The Al₂O₃ thickness of ∼10 nm is determined by two factors; one is the current-voltage (I–V) characteristics of the LEDs which exhibit the highest rectification ratio with moderately lower resistivity. Indeed, thinner Al₂O₃ (<10 nm) causes decrease in the rectification ratio while thicker Al₂O₃ (>10 nm) results in drastic increase in the resistivity. And another factor is the PL characteristics of the ZnO active component which has been reported in the Ref. 20 exhibiting the strongest PL intensity from the ZnO active component at the Al₂O₃ thickness of ∼12 nm. Trimethylaluminum and H₂O were used as Al and O precursors, respectively. Finally, both of Eu-doped ZnO and n-type ZnO were grown by sputtering-assisted metalorganic chemical vapor deposition on a horizontal substrate holder with a resistive heater in a bell-jar chamber. The growth was initiated with the formation of 100 nm-thick Eu-doped ZnO layers via the thermal reaction of diethylzinc (DEZn) in O₂ gas at a substrate temperature of 600 °C. Incorporation of Eu³⁺ ions was achieved by RF magnetron sputtering at 40 W using Eu₂O₃ sintered targets (5 cm diameter) that were fixed above the substrate at a 90° angle and at a distance of 12.5 cm. The DEZn and O₂ flow rates were fixed at 1.11 μmol s⁻¹ and 22.2 μmol s⁻¹, respectively, corresponding to a VI/II ratio of 20. This growth was then followed by the formation of 85 nm-thick n-type ZnO films at a substrate temperature of 700 °C. The structural properties were characterized by scanning electron microscopy (SEM) and transmission electron microscopy (TEM: JEM-ARM200F, JEOL). The specimen for the cross-sectional TEM observation was cleaved from the as-grown LED structures into 3 mm × 1 mm × ∼430 μm sizes using a diamond saw, and polished until the 1 mm-thick specimen becomes the 100 μm-thickness with a mechanical polisher using the diamond papers (abrasive grit size: 30 and 9 μm). In addition, thinning of the 100 μm-thick specimen was performed by Argon ion milling until TEM observation area was less than ∼ several 10 nm-thickness. Then the specimen was removed from the milling and mounted on a reinforcement ring for TEM observation. The concentration of Eu in ZnO:Eu was estimated to be 0.1 at% whose details are discussed in our previous publications, ^24–28 as well as Appendix A. After the growth, samples were annealed at 600 °C for 30 min in O₂ ambient to enhance the luminescence from Eu³⁺ ions. Such an enhancement occurs likely due to the improvement of the crystal quality of the ZnO host induced by suppression of oxygen vacancies, which act as deep donors. For comparison, 200 nm-thick n-type ZnO layer, without doping Eu, was grown on both p-GaN templates with and without Al₂O₃ on the top of the template. Two different kinds of optical characterization, indirect and direct excitation, of the fabricated LEDs were carried out by using a conventional photoluminescence (PL) setup utilizing the spectrometer (SpectraPro-2300i, Acton Research Corporation) equipped with Si photodiode arrays (PIXIS 256, Princeton Instruments). The former involves the excitation through the ZnO host excited by a He-Cd laser (IK3452R-F, Kimmon Koha Co. Ltd). In this case, electron hole pairs are generated in the ZnO host, and these carriers are quickly captured at trap levels. Next, an energy transfer takes place from Eu-related trap levels which excites the Eu ions. Finally, Eu-related luminescence is observed originating from the transition of the ⁵D₀ to the ⁷F₂ state. This indirect excitation usually pumps all the transitions, and involves several transition processes such as excitation, relaxation and energy transfer. ⁷ The PL intensity from Eu³⁺ ions in ZnO depends on both the numbers of Eu ions and the efficiency of the energy transfer, which is dependent on the details of the "local environment of the Eu³⁺ ions." ^29–31 By contrast, the latter involves the resonant excitation of Eu³⁺ ions themselves without pumping the ZnO host and hence other factors including the energy transfer efficiency don't need to be considered. In general, an ideal isolated Eu³⁺ ion exhibits a single sharp optical emission associated with radiative transitions within the 4 f manifold where the intra-4f transitions are parity forbidden according to the Laporte rule. However, once such Eu³⁺ ions are incorporated into ZnO having various local environments, a reduction of the symmetry in their local environment relieves this selection rule. Furthermore, this variation of local environments of incorporation centers, for example a neighboring oxygen vacancy, affects the spectral location of the Eu³⁺ ions. Different incorporation centers can be identified by direct excitation of the Eu³⁺ ions with different excitation wavelengths which is usually performed by a tunable laser, in this manuscript, a dye laser (Matisse DRH-W, Spectra Physics). This technique is called combined excitation emission spectroscopy (CEES), and is useful for quantifying the number and relative abundancies of Eu incorporation centers in the ZnO host. ³² Since a tunable dye laser directly excites Eu³⁺ ions to a higher energy state, the PL intensity from Eu ions excited in this way more closely reflects the numbers of Eu³⁺ ions for each center. The I–V characteristics were measured using a power supply device with maximum applied voltage of ±30 V (ADC 6241 A, ADC Corporation). The EL spectrum was obtained by utilizing the same spectrometer with the Si CCD as those of the PL characterization, under current injection of a source measuring unit (B2901A, Agilent Technologies).

Zoom In Zoom Out Reset image size

The TEM image of the fabricated LED structure is shown in Fig. 1b, which exhibits a clear contrast from the ∼10 nm-thick Al₂O₃ in the vicinity of the interface between ZnO:Eu and the p-GaN layer. The closer TEM observation of the interface in Fig. 1c indicated by the dashed line of Fig. 1b reveals that the GaN matrix is composed of a single crystal while the Al₂O₃ layer is amorphous. By contrast, a large number of grain boundaries are observed from the ZnO:Eu layer, suggesting that ZnO is deposited in a grain configuration. In order to elucidate that the carrier recombination predominantly takes place at the ZnO layer, p-GaN/Al₂O₃/n-ZnO heterojunction LEDs are fabricated without doping Eu at the growth of the ZnO active region. The p-GaN/Al₂O₃/n-ZnO heterojunction LEDs exhibits good rectification behavior (see Appendix B) from the results of I–V measurement. Figure 2a shows the EL spectrum of the p-GaN/Al₂O₃/n-ZnO heterojunction LEDs, along with the LED structures without the Al₂O₃ layer for comparison. Light emission is observed at 389 nm from the LEDs with the Al₂O₃ layer, while the EL peak is located at 430 nm from the LEDs without the Al₂O₃ layer. The PL characterization of both p-GaN and n-ZnO are performed in order to clarify the origin of these peaks. The PL spectra of p-GaN and n-ZnO at room temperature (RT) are shown in Fig. 2b. The PL spectrum of the p-GaN exhibits the emission peak at 425 nm, while that of the n-ZnO exhibits the emission peak at 381 nm. Therefore, as a result of the comparison of the PL and EL results, observed EL peak of the LEDs with the Al₂O₃ originates from the n-ZnO, and that of the LEDs without the Al₂O₃ originates from the p-GaN layer. It is noted here that the EL peak is longer than the PL peak, which is likely due to Joule heat. ²¹ These results indicate that the Al₂O₃ layer can act as the EBL of the p-GaN/Al₂O₃/n-ZnO heterojunction LEDs.

Zoom In Zoom Out Reset image size

Next, the device characterization of the LEDs with Eu doping to the ZnO active layer is performed. Figure 3a shows the typical RT PL spectrum of the ZnO:Eu layer embedded in the LED structure. The band edge light emission from GaN and ZnO are observed at ∼430 and ∼390 nm, respectively, while a very broad emission centered at 600–700 nm is also observed, which originates predominantly from oxygen vacancies especially on the surface of ZnO NWs. There are multiple sharp luminescence peaks from Eu³⁺ ions embedded in the ZnO. It is noted here that the red luminescence with the strongest PL intensity from the ⁵D₀–⁷F₂ transition is located between 610 and 625 nm. Several other peaks from Eu ions are also observed at 540 nm (⁵D₁–⁷F₁), 590 nm (⁵D₀–⁷F₁), and 660 nm (⁵D₀–⁷F₃). Figure 3b shows the RT EL spectra of the LED structure with and without the Al₂O₃ EBL under forward and reversed biases (30 and −30 V) at the injection current of 200 and −100 μA, respectively. In the forward bias, the band edge light emission from the ZnO is observed from the LED structures with the EBL, while very broad emission peaked at the red regime is observed from the LED structures without the EBL, which is likely originated from the yellow luminescence of the GaN layer. In both cases, no Eu luminescence is observable under the forward bias. By contrast, Eu luminescence is observed at the peak wavelength of ∼615 nm under the reverse bias, while no Eu luminescence is observable from the LEDs without the EBL. In order to clarify the behavior of such Eu luminescence, both of the PL and EL spectra in the vicinity of Eu luminescence are compared. Figure 3c shows the enlarged PL and EL spectra of the Figs. 3a and 3b, respectively in the vicinity of the ⁵D₀→⁷F₂ transition. The luminescence peaks are located predominantly at 613.4, 619.4 and 622.1 nm under the PL characterization, while the Eu luminescence peaks are observed at mainly 616.7 and 629.6 nm under the EL characterization with reversed bias which are totally different from those under PL characterization. In general, the PL characterization with an excitation by a He–Cd laser exhibits multiple Eu luminescence by indirect excitation of the ZnO host through the energy transfer process. By contrast, the luminescence mechanism under current injection of RE-doped semiconductors can be classified into two types: one is the indirect excitation, in which the energy caused by the recombination of electron-hole pairs in the host material is transferred to RE ions by a non-radiative process, and luminescence occurs by relaxing the electrons from excited states to the ground state. The another is a collisional excitation, where hot electrons with a large momentum collide with RE ions, and hence directly excite the electrons in the RE ions from a lower state to a higher one. Although the details are not known at the current stage, it is thought that a portion of the electrons accumulated near the interface between Al₂O₃ and p-GaN overcome the barrier and hot electrons are generated during the injection process into the ZnO layer when a reverse current is applied to this structure, which results in collisional excitation of the RE ions. Details of such excitation mechanism are described in Appendix C. Indeed, the EL spectrum under the reversed bias voltage does not exhibit light emission from the ZnO host at all. Therefore, observed difference of luminescence spectra under the PL and EL characterization likely originates from different luminescence mechanism under indirect and direct excitation owing to different local environment structure.

Zoom In Zoom Out Reset image size

In order to gain more information about the difference of such luminescence mechanism especially under current injection, the CEES characterization is performed at 10 K for the LED sample with Al₂O₃. First, the luminescence centers which show up under indirect excitation are identified. Figure 4a shows the CEES map of the LED sample. Among all the 18 luminescence centers from Eu³⁺ ions in the ZnO:Eu directly grown on sapphire substrates as a reference (see Appendix D), there are 3 major luminescence centers contributing to the Eu luminescence at the excitation wavelength of λ_ex = 584.2, 582.9 and 574.15 nm, which are assigned as Eu1, Eu2 and Eu18, respectively. Moreover, besides those peaks, two additional luminescence centers predominantly emerges in the vicinity of the luminescence centers, Eu4, Eu5 and Eu6, at the excitation wavelength of λ_ex = 581.7 and 581.1 nm assigned as EuX and EuY, respectively, which are not observed from the reference ZnO:Eu. Here, the peak wavelength and spectral shape of each luminescence center are compared by extracting the PL spectra from the obtained CEES mapping in Fig. 4a. Figure 4b shows the PL spectra of three main luminescence centers of Eu2, EuY and Eu18. By comparing with the RT PL spectrum with an indirect excitation by He-Cd lasers shown at the bottom of Fig. 4b, Eu luminescence at 622.1, 619.4 and 613.4 nm under indirect excitation is obtained mainly from the luminescence centers, Eu2/EuY, Eu2 and Eu18, respectively. It is noted here that the peaks of Eu luminescence do not shift against the measurement temperature although the PL intensity gradually degrades (see Appendix E), suggesting that the peak wavelengths of Eu luminescence at both RT and 10 K (in Fig. 4b) are almost equivalent. The Eu luminescence at 617 nm in the PL spectrum of the luminescence center, EuY, originates from the shoulder of the strong luminescence center of EuX. These results indicate that luminescence centers of Eu2, Eu18 and EuY likely contribute to Eu luminescence under indirect excitation through the energy transfer from the ZnO host. By contrast, when comparing the EL spectrum under reversed bias, collisional excitation, with the PL spectra of various luminescence centers, the shape of the PL spectrum of the luminescence center of Eu1 located at the longest excitation wavelength is similar to that of the EL spectrum, as shown in Fig. 4c. Therefore, the luminescence center of Eu1 likely contributes to Eu luminescence under collisional excitation of the LEDs. These series of CEES characterization as well as comparison with the PL/EL spectra would contribute to the elucidation of the mechanism of Eu luminescence and how to achieve the Eu luminescence under a forward/reverse bias. Indeed, the abundancy of Eu luminescence centers, Eu2, EuY and Eu18 would need to increase if increasing the brightness of Eu luminescence under forward bias and the Eu1 for Eu luminescence under reversed bias.

Zoom In Zoom Out Reset image size

Heterojunction p-GaN/n-ZnO LED structure using ZnO:Eu as the active layer is proposed and demonstrated in order to realize a low-cost and environmentally friendly red LEDs. Chemically stable Al₂O₃ is inserted as an EBL between p-GaN and ZnO:Eu/n-ZnO to facilitate the injection of carriers into the ZnO:Eu active layer. Moderate thickness of Al₂O₃ (∼10 nm) in p-GaN/Al₂O₃/n-ZnO heterojunction LEDs enables the carrier recombination at the ZnO layer. Device characteristics of the p-GaN/Al₂O₃/ZnO:Eu/n-ZnO LED structures shows Eu luminescence under current injection with reversed bias voltage. Detailed optical characteristics of the ZnO:Eu layer in the LED structures utilizing the CEES measurement enables the identification of the luminescence center contributing to Eu luminescence under both indirect excitation and collisional excitation. Indeed, the luminescence center contributing to Eu luminescence under indirect excitation via the ZnO host is different from that under collisional excitation. These results would enable to understand the Eu luminescence mechanism in ZnO:Eu, resulting in the realization of high-brightness LED structures based on rare-Earth doped ZnO as an active component.

This work was partly supported by Grant-in-Aid for Scientific Research (B) (Grant No. 19H02573), Grant-in-Aid for Specially Promoted Research (Grant No. 18H05212) of the Japan Society for the Promotion of Science, Nanotechnology Platform (Grant No. JPMXP09F21OS0021 and JPMXP09S21OS0013) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the Murata Science Foundation, the Samco Foundation, the Canon Foundation and Iketani Science and Technology Foundation.

Our previous publication reveals that the concentration of Eu in ZnO:Eu is estimated to be 0.1 at% by energy dispersive X-ray spectroscopy (EDS: JEOL Ltd., EX-64175JMV) for both of secondary ion mass spectroscopy or X-ray fluorescence. ^25,26 In this manuscript, the Eu concentration of ZnO:Eu is characterized by the EDS equipped in the TEM facility using the specimen prepared for the TEM observation as shown in Figs. A·1a. Figure A·1b shows the Eu concentration profile (at%) as a function of the distance from the p-GaN/Al₂O₃ interface. The concentration profile shows clear increase in the Eu concentration at the ZnO:Eu layer. The difference of the Eu concentration between ZnO:Eu and n-ZnO, p-GaN layers is approximately ∼0.12 at% which is almost the same order of the lower detection limit of ∼0.1 at%. The observed background of Eu concentration (∼0.1 at%) at the n-ZnO and p-GaN layers compared with the Al₂O₃ layer is due to unintentional doping of Eu during the growth of both ZnO and GaN. This is because our growth facilities often deal with Eu containing materials, leading to the contamination of Eu caused by the memory effect of Eu adatoms. Therefore, at this moment, the Eu concentration cannot be precisely quantified, but it can be concluded at least that there likely exist Eu ions in the ZnO:Eu layer whose concentration is approximately at the order of 0.1 at%, which is consistent with our previous publications.

Zoom In Zoom Out Reset image size

I-V measurement is a method of measuring the current value flowing through a sample while changing the voltage. In the case of an ideal light emitting diode, the I–V characteristics between the p(n) and the p(n)-type semiconductors exhibit ohmic properties if p(n)-type dopants are properly doped into the semiconductor host, and the I–V characteristics between the p and n-type semiconductor show rectification behavior. Figure A·2a shows the I-V characteristics between p-GaN–p-GaN, n-ZnO–n-ZnO, which clearly exhibit the ohmic contact, indicating that both of p- and n- dopants are properly doped into GaN and ZnO hosts, respectively. Figure A·2b shows the I-V characteristics of fabricated p-GaN/Al₂O₃/ZnO:Eu/n-ZnO heterojunction LEDs with and without the Al₂O₃ EBL, both of which exhibit the rectification behavior with a turn on voltage of ∼4 and ∼7 V for the LEDs, respectively. Higher turn on of the LED with the EBL is owing to the existence of Al₂O₃ layer.

Zoom In Zoom Out Reset image size

Figure A·3 shows the schematic model of the mechanism of the collisional excitation of Eu³⁺ ions in the ZnO:Eu layers of the fabricated LEDs under the reversed bias. The following behavior is thought to occur when a reverse voltage is applied to the fabricated LEDs:

• -  
  Due to the application of a large reverse voltage, electrons from the p-GaN accumulate in the vicinity of the interface between Al₂O₃ and p-GaN which results in a population inversion.
• -  
  Accumulated electrons tunnel into ZnO:Eu as hot electrons through the Al₂O₃ barrier layer owing to their strong applied electric field.
• -  
  A portion of the hot electrons collide with Eu³⁺ ions in the diffusion process through the ZnO:Eu layer, and the kinetic energy of the hot electrons directly transfer to Eu³⁺ ions to excite the electrons to the excited state. Finally luminescence from Eu³⁺ ions takes place by relaxing the electrons from the excited to the ground state.

Zoom In Zoom Out Reset image size

In the ZnO:Eu, luminescence from both Eu³⁺ ions and the ZnO host is observed under indirect excitation while only luminescence from Eu³⁺ ions is observable under direct excitation. Therefore, obtained EL spectra under the forward bias originate from the luminescence centers contributing to Eu luminescence under indirect excitation via the ZnO host, while EL spectra under the reversed bias originate from those under direct excitation due to the collisional excitation. Further studies including the dependence of the EL intensity of Eu luminescence on the injection current are required to elucidate the mechanism of hot carrier generation, ^33,34 and will be a subject of future publication.

In the CEES measurement, the PL spectra are collected by changing the excitation wavelength using a tunable laser (in this manuscript, dye lasers), and Eu luminescence center is excited with a photon energy corresponding exactly to the ⁷F₀–⁵D₀ transition of Eu³⁺ ions. Two-dimensional mapping of excitation wavelength vs PL spectrum is obtained by observing the PL spectrum in the vicinity of the ⁵D₀–⁷F₂ transition. At this stage, since the ⁷F₀–⁵D₀ transition does not split due to the crystal field, different luminescence centers originate for different excitation energies. Therefore, it is important to investigate the luminescence mechanism of Eu ions in order to increase the intensity of Eu luminescence.

In this contribution, Eu luminescence from the LED structures is observable under current injection with reversed bias voltage due to collisional excitation. Therefore, CEES measurement is performed to analyze and identify the luminescence centers under current injection. Two types of the samples are characterized: one is the LED structure, p-GaN/Al₂O₃/ZnO:Eu/n-ZnO, and another is the ZnO:Eu layer directly grown on sapphire substrates as reference. From the CEES mapping, the ratio of each luminescence centers is totally different between the two structures. Figure A·4a shows the CEES mapping of the ZnO:Eu/sapphire which exhibits 18 different luminescence centers depending on the excitation wavelength which is owing to different local environmental structures. Here, Eu1, 2, and 3 are assigned in order from the luminescence center at the longer excitation wavelength. By contrast, from the CEES mapping of the LED structure shown in Fig. A·4b, among these 18 peaks from the reference sample, luminescence centers assigned as Eu1, 2, 8, 11, 12, 13, 15, 16, and 18 are observed. Furthermore, two additional distinct peaks are observed at slightly different luminescence wavelength in the vicinity of Eu 4–6. These peaks are assigned as EuX and EuY at the excitation wavelength of λ_ex = 581.7 and 581.1 nm, respectively. This difference is considered to be due to the different strains and defects of the ZnO host in the ZnO:Eu layer owing to different lattice constants of Al₂O₃ as substrates. Such presence of a large number of luminescence centers in the ZnO:Eu film configuration has been already reported, and it would be expected to become a stepping stone to elucidate the Eu luminescence mechanism in the ZnO:Eu.

Zoom In Zoom Out Reset image size

Rare-Earth ions, including Eu³⁺ ions, are characterized by partially-filled 4f shells, which are localized inside of completely filled 5 s and 5p shells. This localization shields the electrons in the 4 f shell from the surrounding environment. As a result, rare-Earth ions maintain their "atom-like" properties and exhibit sharp optical luminescence/absorption bands associated with radiative transitions within the 4 f manifold, which are virtually insensitive to surrounding environment including injection current and temperature. Figure A·5a shows the temperature dependence of the PL spectra in the vicinity of ∼613 nm from Eu³⁺ ions in the LEDs. The peak wavelength of the PL spectrum at 15 K is located at 613.2 nm, exhibiting the gradual redshift with increasing measurement temperature. The peak wavelength becomes 613.7 nm at 300 K, corresponding to the average shift of the peak wavelength of <0.002 nm K⁻¹. Figure A·5b shows the temperature dependence of the peak intensity of Eu luminescence at ∼613 nm whose spectra are shown in Fig. A·5a. All PL intensities are normalized by those measured at 15 K. The PL intensity shows gradual decrease with increased temperature, and the normalized PL intensity is almost 0.2 at 300 K.

Zoom In Zoom Out Reset image size